Refactor layers. (#1866)

# What does this PR do?

<!--
Congratulations! You've made it this far! You're not quite done yet
though.

Once merged, your PR is going to appear in the release notes with the
title you set, so make sure it's a great title that fully reflects the
extent of your awesome contribution.

Then, please replace this with a description of the change and which
issue is fixed (if applicable). Please also include relevant motivation
and context. List any dependencies (if any) that are required for this
change.

Once you're done, someone will review your PR shortly (see the section
"Who can review?" below to tag some potential reviewers). They may
suggest changes to make the code even better. If no one reviewed your PR
after a week has passed, don't hesitate to post a new comment
@-mentioning the same persons---sometimes notifications get lost.
-->

<!-- Remove if not applicable -->

Fixes # (issue)


## Before submitting
- [ ] This PR fixes a typo or improves the docs (you can dismiss the
other checks if that's the case).
- [ ] Did you read the [contributor
guideline](https://github.com/huggingface/transformers/blob/main/CONTRIBUTING.md#start-contributing-pull-requests),
      Pull Request section?
- [ ] Was this discussed/approved via a Github issue or the
[forum](https://discuss.huggingface.co/)? Please add a link
      to it if that's the case.
- [ ] Did you make sure to update the documentation with your changes?
Here are the
[documentation
guidelines](https://github.com/huggingface/transformers/tree/main/docs),
and
[here are tips on formatting
docstrings](https://github.com/huggingface/transformers/tree/main/docs#writing-source-documentation).
- [ ] Did you write any new necessary tests?


## Who can review?

Anyone in the community is free to review the PR once the tests have
passed. Feel free to tag
members/contributors who may be interested in your PR.

<!-- Your PR will be replied to more quickly if you can figure out the
right person to tag with @


@OlivierDehaene OR @Narsil

 -->

server/tests/utils/test_layers.py

@@ -1,5 +1,5 @@
 import torch
-from text_generation_server.utils.layers import (
+from text_generation_server.layers import (
     TensorParallelEmbedding,
 )
 

server/text_generation_server/layers/__init__.py

@@ -0,0 +1,14 @@
+from text_generation_server.layers.tensor_parallel import (
+    TensorParallelColumnLinear,
+    TensorParallelRowLinear,
+    TensorParallelEmbedding,
+)
+from text_generation_server.layers.speculative import SpeculativeHead
+from text_generation_server.layers.linear import (
+    get_linear,
+    FastLinear,
+)
+
+# Just to add the `load` methods.
+from text_generation_server.layers.layernorm import load_layer_norm
+from text_generation_server.layers.conv import load_conv2d

server/text_generation_server/layers/bnb.py

@@ -0,0 +1,106 @@
+import torch
+from loguru import logger
+from functools import lru_cache
+import bitsandbytes as bnb
+from bitsandbytes.nn import Int8Params, Params4bit
+
+
+@lru_cache(1)
+def warn_deprecate_bnb():
+    logger.warning(
+        "Bitsandbytes 8bit is deprecated, using `eetq` is a drop-in replacement, and has much better performnce"
+    )
+
+
+class Linear8bitLt(torch.nn.Module):
+    def __init__(
+        self,
+        weight,
+        bias,
+        has_fp16_weights=True,
+        memory_efficient_backward=False,
+        threshold=0.0,
+        index=None,
+    ):
+        super().__init__()
+        assert (
+            not memory_efficient_backward
+        ), "memory_efficient_backward is no longer required and the argument is deprecated in 0.37.0 and will be removed in 0.39.0"
+        self.state = bnb.MatmulLtState()
+        self.index = index
+
+        # Necessary for stacked layers
+        self.state.threshold = threshold
+        self.state.has_fp16_weights = has_fp16_weights
+        self.state.memory_efficient_backward = memory_efficient_backward
+        if threshold > 0.0 and not has_fp16_weights:
+            self.state.use_pool = True
+
+        self.weight = Int8Params(
+            weight.data,
+            has_fp16_weights=has_fp16_weights,
+            requires_grad=has_fp16_weights,
+        )
+        self.weight.cuda(weight.device)
+        self.bias = bias
+
+    def init_8bit_state(self):
+        self.state.CB = self.weight.CB
+        self.state.SCB = self.weight.SCB
+        self.weight.CB = None
+        self.weight.SCB = None
+
+    def forward(self, x: torch.Tensor):
+        self.state.is_training = self.training
+        if self.weight.CB is not None:
+            self.init_8bit_state()
+
+        # weights are cast automatically as Int8Params, but the bias has to be cast manually
+        if self.bias is not None and self.bias.dtype != x.dtype:
+            self.bias.data = self.bias.data.to(x.dtype)
+
+        out = bnb.matmul(x, self.weight, bias=self.bias, state=self.state)
+
+        if not self.state.has_fp16_weights:
+            if self.state.CB is not None and self.state.CxB is not None:
+                # we converted 8-bit row major to turing/ampere format in the first inference pass
+                # we no longer need the row-major weight
+                del self.state.CB
+                self.weight.data = self.state.CxB
+        return out
+
+
+class Linear4bit(nn.Module):
+    def __init__(self, weight, bias, quant_type):
+        super().__init__()
+        self.weight = Params4bit(
+            weight.data,
+            requires_grad=False,
+            compress_statistics=True,
+            quant_type=quant_type,
+        )
+        self.compute_dtype = None
+        self.weight.cuda(weight.device)
+        self.bias = bias
+
+    def forward(self, x: torch.Tensor):
+        # weights are cast automatically as Int8Params, but the bias has to be cast manually
+        if self.bias is not None and self.bias.dtype != x.dtype:
+            self.bias.data = self.bias.data.to(x.dtype)
+
+        if getattr(self.weight, "quant_state", None) is None:
+            print(
+                "FP4 quantization state not initialized. Please call .cuda() or .to(device) on the LinearFP4 layer first."
+            )
+        inp_dtype = x.dtype
+        if self.compute_dtype is not None:
+            x = x.to(self.compute_dtype)
+
+        bias = None if self.bias is None else self.bias.to(self.compute_dtype)
+        out = bnb.matmul_4bit(
+            x, self.weight.t(), bias=bias, quant_state=self.weight.quant_state
+        )
+
+        out = out.to(inp_dtype)
+
+        return out

server/text_generation_server/layers/conv.py

@@ -0,0 +1,41 @@
+from accelerate import init_empty_weights
+import torch
+
+
+@classmethod
+def load_conv2d(cls, prefix, weights, in_channels, out_channels, kernel_size, stride):
+    weight = weights.get_tensor(f"{prefix}.weight")
+    bias = weights.get_tensor(f"{prefix}.bias")
+    with init_empty_weights():
+        conv2d = cls(
+            in_channels=in_channels,
+            out_channels=out_channels,
+            kernel_size=kernel_size,
+            stride=stride,
+        )
+
+    conv2d.weight = torch.nn.Parameter(weight)
+    conv2d.bias = torch.nn.Parameter(bias)
+    return conv2d
+
+
+@classmethod
+def load_conv2d_no_bias(
+    cls, prefix, weights, in_channels, out_channels, kernel_size, stride
+):
+    weight = weights.get_tensor(f"{prefix}.weight")
+    with init_empty_weights():
+        conv2d = cls(
+            in_channels=in_channels,
+            out_channels=out_channels,
+            kernel_size=kernel_size,
+            stride=stride,
+        )
+
+    conv2d.weight = torch.nn.Parameter(weight)
+    conv2d.bias = None
+    return conv2d
+
+
+torch.nn.Conv2d.load = load_conv2d
+torch.nn.Conv2d.load_no_bias = load_conv2d_no_bias

server/text_generation_server/layers/eetq.py

@@ -0,0 +1,25 @@
+import torch
+from EETQ import quant_weights, w8_a16_gemm
+
+
+class EETQLinear(torch.nn.Module):
+    def __init__(
+        self,
+        weight,
+        bias,
+    ) -> None:
+        super().__init__()
+        device = weight.device
+        if weight.dtype != torch.float16:
+            weight = weight.to(dtype=torch.float16)
+        weight = torch.t(weight).contiguous().cpu()
+        weight, scale = quant_weights(weight, torch.int8, False)
+
+        self.weight = weight.cuda(device)
+        self.scale = scale.cuda(device)
+        self.bias = bias.cuda(device) if bias is not None else None
+
+    def forward(self, input: torch.Tensor) -> torch.Tensor:
+        output = w8_a16_gemm(input, self.weight, self.scale)
+        output = output + self.bias if self.bias is not None else output
+        return output

server/text_generation_server/layers/fp8.py

@@ -0,0 +1,43 @@
+import torch
+
+
+def fp8_quantize(weight, qdtype=torch.float8_e4m3fn):
+    device = weight.device
+    # weight, scale = quant_weights(weight, torch.int8, False)
+    finfo = torch.finfo(qdtype)
+    # Calculate the scale as dtype max divided by absmax
+    scale = finfo.max / weight.abs().max().clamp(min=1e-12)
+    # scale and clamp the tensor to bring it to
+    # the representative range of float8 data type
+    # (as default cast is unsaturated)
+    qweight = (weight * scale).clamp(min=finfo.min, max=finfo.max)
+    # Return both float8 data and the inverse scale (as float),
+    # as both required as inputs to torch._scaled_mm
+    qweight = qweight.to(qdtype)
+    scale = scale.float().reciprocal()
+    return qweight, scale
+
+
+class Fp8Linear(torch.nn.Module):
+    def __init__(
+        self,
+        weight,
+        bias,
+    ) -> None:
+        super().__init__()
+        self.dtype = weight.dtype
+        self.qweight, self.scale = fp8_quantize(weight)
+
+        self.bias = bias if bias is not None else None
+
+    def forward(self, input: torch.Tensor) -> torch.Tensor:
+        qinput, scale = fp8_quantize(input)
+        output, _ = torch._scaled_mm(
+            qinput,
+            self.qweight.t(),
+            out_dtype=self.dtype,
+            scale_a=scale,
+            scale_b=self.scale,
+            bias=self.bias,
+        )
+        return output

server/text_generation_server/layers/gptq/__init__.py

@@ -0,0 +1,39 @@
+import os
+import torch
+from text_generation_server.utils.import_utils import (
+    SYSTEM,
+)
+
+try:
+    major, _minor = torch.cuda.get_device_capability()
+except Exception:
+    major = 1
+
+HAS_EXLLAMA = False
+CAN_EXLLAMA = major >= 8 or SYSTEM == "rocm"
+V2 = os.getenv("EXLLAMA_VERSION", "2") == "2"
+if os.getenv("DISABLE_EXLLAMA") == "True":
+    HAS_EXLLAMA = False
+elif CAN_EXLLAMA:
+    try:
+        if V2:
+            from text_generation_server.layers.gptq.exllamav2 import (
+                QuantLinear as ExllamaQuantLinear,
+                create_exllama_buffers,
+                set_device,
+            )
+
+            HAS_EXLLAMA = "2"
+        else:
+            from text_generation_server.layers.gptq.exllama import (
+                Ex4bitLinear as ExllamaQuantLinear,
+                create_exllama_buffers,
+                set_device,
+            )
+
+            HAS_EXLLAMA = "1"
+
+    except ImportError:
+        pass
+
+from text_generation_server.layers.gptq.quant_linear import QuantLinear

server/text_generation_server/layers/gptq/exllamav2.py

@@ -119,6 +119,8 @@ def ext_make_q_matrix(w: dict, temp_dq, key: str = None):
                 none_tensor,
                 temp_dq,
             )
+    else:
+        RuntimeError("Cannot create handle")
 
 
 DEVICE = None

server/text_generation_server/layers/gptq/quant_linear.py

@@ -0,0 +1,356 @@
+import math
+import numpy as np
+import torch
+import torch.nn as nn
+from torch.cuda.amp import custom_fwd
+
+import triton
+import triton.language as tl
+from . import custom_autotune
+
+
+# code based https://github.com/fpgaminer/GPTQ-triton
+@custom_autotune.autotune(
+    configs=[
+        triton.Config(
+            {
+                "BLOCK_SIZE_M": 64,
+                "BLOCK_SIZE_N": 256,
+                "BLOCK_SIZE_K": 32,
+                "GROUP_SIZE_M": 8,
+            },
+            num_stages=4,
+            num_warps=4,
+        ),
+        triton.Config(
+            {
+                "BLOCK_SIZE_M": 128,
+                "BLOCK_SIZE_N": 128,
+                "BLOCK_SIZE_K": 32,
+                "GROUP_SIZE_M": 8,
+            },
+            num_stages=4,
+            num_warps=4,
+        ),
+        triton.Config(
+            {
+                "BLOCK_SIZE_M": 64,
+                "BLOCK_SIZE_N": 128,
+                "BLOCK_SIZE_K": 32,
+                "GROUP_SIZE_M": 8,
+            },
+            num_stages=4,
+            num_warps=4,
+        ),
+        triton.Config(
+            {
+                "BLOCK_SIZE_M": 128,
+                "BLOCK_SIZE_N": 32,
+                "BLOCK_SIZE_K": 32,
+                "GROUP_SIZE_M": 8,
+            },
+            num_stages=4,
+            num_warps=4,
+        ),
+        triton.Config(
+            {
+                "BLOCK_SIZE_M": 64,
+                "BLOCK_SIZE_N": 64,
+                "BLOCK_SIZE_K": 32,
+                "GROUP_SIZE_M": 8,
+            },
+            num_stages=4,
+            num_warps=4,
+        ),
+        triton.Config(
+            {
+                "BLOCK_SIZE_M": 64,
+                "BLOCK_SIZE_N": 128,
+                "BLOCK_SIZE_K": 32,
+                "GROUP_SIZE_M": 8,
+            },
+            num_stages=2,
+            num_warps=8,
+        ),
+        triton.Config(
+            {
+                "BLOCK_SIZE_M": 64,
+                "BLOCK_SIZE_N": 64,
+                "BLOCK_SIZE_K": 64,
+                "GROUP_SIZE_M": 8,
+            },
+            num_stages=3,
+            num_warps=8,
+        ),
+        triton.Config(
+            {
+                "BLOCK_SIZE_M": 32,
+                "BLOCK_SIZE_N": 32,
+                "BLOCK_SIZE_K": 128,
+                "GROUP_SIZE_M": 8,
+            },
+            num_stages=2,
+            num_warps=4,
+        ),
+    ],
+    key=["M", "N", "K"],
+    nearest_power_of_two=True,
+    prune_configs_by={
+        "early_config_prune": custom_autotune.matmul248_kernel_config_pruner,
+        "perf_model": None,
+        "top_k": None,
+    },
+)
+@triton.jit
+def matmul_248_kernel(
+    a_ptr,
+    b_ptr,
+    c_ptr,
+    scales_ptr,
+    zeros_ptr,
+    g_ptr,
+    M,
+    N,
+    K,
+    bits,
+    maxq,
+    stride_am,
+    stride_ak,
+    stride_bk,
+    stride_bn,
+    stride_cm,
+    stride_cn,
+    stride_scales,
+    stride_zeros,
+    BLOCK_SIZE_M: tl.constexpr,
+    BLOCK_SIZE_N: tl.constexpr,
+    BLOCK_SIZE_K: tl.constexpr,
+    GROUP_SIZE_M: tl.constexpr,
+):
+    """
+    Compute the matrix multiplication C = A x B.
+    A is of shape (M, K) float16
+    B is of shape (K//8, N) int32
+    C is of shape (M, N) float16
+    scales is of shape (G, N) float16
+    zeros is of shape (G, N) float16
+    g_ptr is of shape (K) int32
+    """
+    infearure_per_bits = 32 // bits
+
+    pid = tl.program_id(axis=0)
+    num_pid_m = tl.cdiv(M, BLOCK_SIZE_M)
+    num_pid_n = tl.cdiv(N, BLOCK_SIZE_N)
+    num_pid_k = tl.cdiv(K, BLOCK_SIZE_K)
+    num_pid_in_group = GROUP_SIZE_M * num_pid_n
+    group_id = pid // num_pid_in_group
+    first_pid_m = group_id * GROUP_SIZE_M
+    group_size_m = min(num_pid_m - first_pid_m, GROUP_SIZE_M)
+    pid_m = first_pid_m + (pid % group_size_m)
+    pid_n = (pid % num_pid_in_group) // group_size_m
+
+    offs_am = pid_m * BLOCK_SIZE_M + tl.arange(0, BLOCK_SIZE_M)
+    offs_bn = pid_n * BLOCK_SIZE_N + tl.arange(0, BLOCK_SIZE_N)
+    offs_k = tl.arange(0, BLOCK_SIZE_K)
+    a_ptrs = a_ptr + (
+        offs_am[:, None] * stride_am + offs_k[None, :] * stride_ak
+    )  # (BLOCK_SIZE_M, BLOCK_SIZE_K)
+    a_mask = offs_am[:, None] < M
+    # b_ptrs is set up such that it repeats elements along the K axis 8 times
+    b_ptrs = b_ptr + (
+        (offs_k[:, None] // infearure_per_bits) * stride_bk
+        + offs_bn[None, :] * stride_bn
+    )  # (BLOCK_SIZE_K, BLOCK_SIZE_N)
+    g_ptrs = g_ptr + offs_k
+    # shifter is used to extract the N bits of each element in the 32-bit word from B
+    scales_ptrs = scales_ptr + offs_bn[None, :]
+    zeros_ptrs = zeros_ptr + (offs_bn[None, :] // infearure_per_bits)
+
+    shifter = (offs_k % infearure_per_bits) * bits
+    zeros_shifter = (offs_bn % infearure_per_bits) * bits
+    accumulator = tl.zeros((BLOCK_SIZE_M, BLOCK_SIZE_N), dtype=tl.float32)
+
+    for k in range(0, num_pid_k):
+        g_idx = tl.load(g_ptrs)
+
+        # Fetch scales and zeros; these are per-outfeature and thus reused in the inner loop
+        scales = tl.load(
+            scales_ptrs + g_idx[:, None] * stride_scales
+        )  # (BLOCK_SIZE_K, BLOCK_SIZE_N,)
+        zeros = tl.load(
+            zeros_ptrs + g_idx[:, None] * stride_zeros
+        )  # (BLOCK_SIZE_K, BLOCK_SIZE_N,)
+
+        zeros = (zeros >> zeros_shifter[None, :]) & maxq
+        zeros = (zeros + 1) & maxq  # eventually avoid overflow
+
+        a = tl.load(a_ptrs, mask=a_mask, other=0.0)  # (BLOCK_SIZE_M, BLOCK_SIZE_K)
+        b = tl.load(b_ptrs)  # (BLOCK_SIZE_K, BLOCK_SIZE_N), but repeated
+
+        # Now we need to unpack b (which is N-bit values) into 32-bit values
+        b = (b >> shifter[:, None]) & maxq  # Extract the N-bit values
+        b = (b - zeros) * scales  # Scale and shift
+
+        accumulator += tl.dot(a, b)
+        a_ptrs += BLOCK_SIZE_K
+        b_ptrs += (BLOCK_SIZE_K // infearure_per_bits) * stride_bk
+        g_ptrs += BLOCK_SIZE_K
+
+    c_ptrs = c_ptr + stride_cm * offs_am[:, None] + stride_cn * offs_bn[None, :]
+    c_mask = (offs_am[:, None] < M) & (offs_bn[None, :] < N)
+    tl.store(c_ptrs, accumulator, mask=c_mask)
+
+
+def matmul248(input, qweight, scales, qzeros, g_idx, bits, maxq):
+    with torch.cuda.device(input.device):
+        output = torch.empty(
+            (input.shape[0], qweight.shape[1]), device=input.device, dtype=torch.float16
+        )
+        grid = lambda META: (
+            triton.cdiv(input.shape[0], META["BLOCK_SIZE_M"])
+            * triton.cdiv(qweight.shape[1], META["BLOCK_SIZE_N"]),
+        )
+        matmul_248_kernel[grid](
+            input,
+            qweight,
+            output,
+            scales,
+            qzeros,
+            g_idx,
+            input.shape[0],
+            qweight.shape[1],
+            input.shape[1],
+            bits,
+            maxq,
+            input.stride(0),
+            input.stride(1),
+            qweight.stride(0),
+            qweight.stride(1),
+            output.stride(0),
+            output.stride(1),
+            scales.stride(0),
+            qzeros.stride(0),
+        )
+        return output
+
+
+class QuantLinearFunction(torch.autograd.Function):
+    @staticmethod
+    @custom_fwd(cast_inputs=torch.float16)
+    def forward(ctx, input, qweight, scales, qzeros, g_idx, bits, maxq):
+        output = matmul248(input, qweight, scales, qzeros, g_idx, bits, maxq)
+        return output
+
+
+class QuantLinear(nn.Module):
+    def __init__(self, qweight, qzeros, scales, g_idx, bias, bits, groupsize):
+        super().__init__()
+        self.register_buffer("qweight", qweight)
+        self.register_buffer("qzeros", qzeros)
+        self.register_buffer("scales", scales)
+        self.register_buffer("g_idx", g_idx)
+        if bias is not None:
+            self.register_buffer("bias", bias)
+        else:
+            self.bias = None
+        if bits not in [2, 4, 8]:
+            raise NotImplementedError("Only 2,4,8 bits are supported.")
+        self.bits = bits
+        self.maxq = 2**self.bits - 1
+        self.groupsize = groupsize
+
+        self.outfeatures = qweight.shape[1]
+        self.infeatures = qweight.shape[0] * 32 // bits
+
+    @classmethod
+    def new(cls, bits, groupsize, infeatures, outfeatures, bias):
+        if bits not in [2, 4, 8]:
+            raise NotImplementedError("Only 2,4,8 bits are supported.")
+
+        qweight = torch.zeros((infeatures // 32 * bits, outfeatures), dtype=torch.int32)
+        qzeros = torch.zeros(
+            (math.ceil(infeatures / groupsize), outfeatures // 32 * bits),
+            dtype=torch.int32,
+        )
+        scales = torch.zeros(
+            (math.ceil(infeatures / groupsize), outfeatures), dtype=torch.float16
+        )
+        g_idx = torch.tensor(
+            [i // groupsize for i in range(infeatures)], dtype=torch.int32
+        )
+        if bias:
+            bias = torch.zeros((outfeatures), dtype=torch.float16)
+        else:
+            bias = None
+        return cls(qweight, qzeros, scales, g_idx, bias, bits, groupsize)
+
+    def pack(self, linear, scales, zeros, g_idx=None):
+        self.g_idx = g_idx.clone() if g_idx is not None else self.g_idx
+
+        scales = scales.t().contiguous()
+        zeros = zeros.t().contiguous()
+        scale_zeros = zeros * scales
+        self.scales = scales.clone().half()
+        if linear.bias is not None:
+            self.bias = linear.bias.clone().half()
+
+        intweight = []
+        for idx in range(self.infeatures):
+            intweight.append(
+                torch.round(
+                    (linear.weight.data[:, idx] + scale_zeros[self.g_idx[idx]])
+                    / self.scales[self.g_idx[idx]]
+                ).to(torch.int)[:, None]
+            )
+        intweight = torch.cat(intweight, dim=1)
+        intweight = intweight.t().contiguous()
+        intweight = intweight.numpy().astype(np.uint32)
+        qweight = np.zeros(
+            (intweight.shape[0] // 32 * self.bits, intweight.shape[1]), dtype=np.uint32
+        )
+        i = 0
+        row = 0
+        while row < qweight.shape[0]:
+            if self.bits in [2, 4, 8]:
+                for j in range(i, i + (32 // self.bits)):
+                    qweight[row] |= intweight[j] << (self.bits * (j - i))
+                i += 32 // self.bits
+                row += 1
+            else:
+                raise NotImplementedError("Only 2,4,8 bits are supported.")
+
+        qweight = qweight.astype(np.int32)
+        self.qweight = torch.from_numpy(qweight)
+
+        zeros -= 1
+        zeros = zeros.numpy().astype(np.uint32)
+        qzeros = np.zeros(
+            (zeros.shape[0], zeros.shape[1] // 32 * self.bits), dtype=np.uint32
+        )
+        i = 0
+        col = 0
+        while col < qzeros.shape[1]:
+            if self.bits in [2, 4, 8]:
+                for j in range(i, i + (32 // self.bits)):
+                    qzeros[:, col] |= zeros[:, j] << (self.bits * (j - i))
+                i += 32 // self.bits
+                col += 1
+            else:
+                raise NotImplementedError("Only 2,4,8 bits are supported.")
+
+        qzeros = qzeros.astype(np.int32)
+        self.qzeros = torch.from_numpy(qzeros)
+
+    def forward(self, x):
+        out_shape = x.shape[:-1] + (self.outfeatures,)
+        out = QuantLinearFunction.apply(
+            x.reshape(-1, x.shape[-1]),
+            self.qweight,
+            self.scales,
+            self.qzeros,
+            self.g_idx,
+            self.bits,
+            self.maxq,
+        )
+        out = out + self.bias if self.bias is not None else out
+        return out.reshape(out_shape)

server/text_generation_server/layers/layernorm.py

@@ -0,0 +1,185 @@
+import torch
+from torch import nn
+from accelerate import init_empty_weights
+from text_generation_server.utils.import_utils import (
+    SYSTEM,
+)
+
+
+# Monkey patching
+@classmethod
+def load_layer_norm(cls, prefix, weights, eps):
+    weight = weights.get_tensor(f"{prefix}.weight")
+    bias = weights.get_tensor(f"{prefix}.bias")
+    with init_empty_weights():
+        ln = cls(weight.shape, eps=eps)
+
+    ln.weight = torch.nn.Parameter(weight)
+    ln.bias = torch.nn.Parameter(bias)
+    return ln
+
+
+@classmethod
+def load_layer_norm_no_bias(cls, prefix, weights, eps):
+    weight = weights.get_tensor(f"{prefix}.weight")
+    with init_empty_weights():
+        ln = cls(weight.shape, eps=eps)
+
+    ln.weight = torch.nn.Parameter(weight)
+    ln.bias = None
+    return ln
+
+
+torch.nn.LayerNorm.load = load_layer_norm
+torch.nn.LayerNorm.load_no_bias = load_layer_norm_no_bias
+
+if SYSTEM == "cuda":
+    import dropout_layer_norm
+
+    class FastLayerNorm(nn.LayerNorm):
+        def forward(self, hidden_states, residual=None):
+            if hidden_states.shape[-1] > 8192:
+                if residual is not None:
+                    hidden_states += residual
+                residual = hidden_states
+
+                return super(FastLayerNorm, self).forward(hidden_states), residual
+            else:
+                (
+                    normed_hidden_states,
+                    residual,
+                    *rest,
+                ) = dropout_layer_norm.dropout_add_ln_fwd(
+                    hidden_states,
+                    residual,
+                    self.weight,
+                    self.bias,
+                    None,
+                    None,
+                    None,
+                    None,
+                    0.0,
+                    self.eps,
+                    1.0,
+                    0,
+                    None,
+                    False,
+                    False,
+                )
+                if residual is None:
+                    residual = hidden_states
+
+                return normed_hidden_states, residual
+
+elif SYSTEM == "rocm":
+    from vllm import layernorm_ops
+
+    class FastLayerNorm(nn.LayerNorm):
+        def forward(self, hidden_states, residual=None):
+            if residual is not None:
+                hidden_states += residual
+            residual = hidden_states
+
+            return super().forward(hidden_states), residual
+
+elif SYSTEM == "xpu":
+    import intel_extension_for_pytorch as ipex
+
+    class FastLayerNorm(nn.LayerNorm):
+        def forward(self, hidden_states, residual=None):
+            res_out = hidden_states
+            out = ipex.llm.functional.add_layer_norm(
+                residual, hidden_states, self.weight, self.bias, self.eps, True
+            )
+            if residual is not None:
+                res_out = residual
+            return out, res_out
+
+
+class FastRMSNorm(nn.Module):
+    def __init__(self, weight: torch.Tensor, eps: float):
+        super().__init__()
+
+        self.weight = nn.Parameter(weight)
+        self.variance_epsilon = eps
+
+    @classmethod
+    def load(cls, prefix, weights, eps=1e-6):
+        weight = weights.get_tensor(f"{prefix}.weight")
+        return cls(weight, eps)
+
+    def forward(self, hidden_states, residual=None):
+        if SYSTEM == "xpu":
+            residual_out = hidden_states
+            out = ipex.llm.functional.add_rms_norm(
+                residual,
+                hidden_states,
+                self.weight,
+                None,
+                self.variance_epsilon,
+                True,
+            )
+            if residual is not None:
+                residual_out = residual
+            return out, residual_out
+        elif hidden_states.shape[-1] > 8192:
+            if residual is not None:
+                hidden_states += residual
+            residual = hidden_states
+
+            hidden_states = hidden_states.to(torch.float32)
+            variance = hidden_states.pow(2).mean(-1, keepdim=True)
+            hidden_states = hidden_states * torch.rsqrt(
+                variance + self.variance_epsilon
+            )
+
+            # convert into half-precision if necessary
+            if self.weight.dtype in [torch.float16, torch.bfloat16]:
+                hidden_states = hidden_states.to(self.weight.dtype)
+
+            return self.weight * hidden_states, residual
+        elif SYSTEM == "cuda":
+            # faster post attention rms norm
+            (
+                normed_hidden_states,
+                res,
+                *rest,
+            ) = dropout_layer_norm.dropout_add_ln_fwd(
+                hidden_states,
+                residual,
+                self.weight,
+                None,
+                None,
+                None,
+                None,
+                None,
+                0.0,
+                self.variance_epsilon,
+                1.0,
+                0,
+                None,
+                False,
+                True,  # Activate RMSNorm
+            )
+            if res is None:
+                res = hidden_states
+
+            return normed_hidden_states, res
+        elif SYSTEM == "rocm":
+            # We use VLLM RMSNorm kernel that can be compiled for RoCm, instead of Flash Attention ones that can not.
+            if residual is not None:
+                hidden_states += residual
+            residual = hidden_states
+
+            out = torch.empty_like(hidden_states)
+            layernorm_ops.rms_norm(
+                out,
+                hidden_states,
+                self.weight.data,
+                self.variance_epsilon,
+            )
+            return out, residual
+        else:
+            raise ValueError(
+                "Your system seem to be not supported. Please check your install or open an issue at https://github.com/huggingface/text-generation-inference/issues with a clear reproduction."
+            )

server/text_generation_server/layers/linear.py

@@ -0,0 +1,153 @@
+import torch
+from torch.nn import functional as F
+from text_generation_server.utils.import_utils import SYSTEM
+
+
+class FastLinear(torch.nn.Module):
+    def __init__(
+        self,
+        weight,
+        bias,
+    ) -> None:
+        super().__init__()
+        self.weight = torch.nn.Parameter(weight)
+        if bias is not None:
+            self.bias = torch.nn.Parameter(bias)
+        else:
+            self.bias = None
+
+    @classmethod
+    def load(cls, config, prefix: str, weights, bias: bool):
+        weight = weights.get_tensor(f"{prefix}.weight")
+        if bias:
+            bias = weights.get_tensor(f"{prefix}.bias")
+        else:
+            bias = None
+        return cls(weight, bias)
+
+    def forward(self, input: torch.Tensor) -> torch.Tensor:
+        return F.linear(input, self.weight, self.bias)
+
+
+def get_linear(weight, bias, quantize):
+    if quantize is None:
+        linear = FastLinear(weight, bias)
+    elif quantize == "eetq":
+        try:
+            from text_generation_server.layers.eetq import EETQLinear
+
+            linear = EETQLinear(weight, bias)
+        except ImportError:
+            raise ImportError(
+                "Please install EETQ from https://github.com/NetEase-FuXi/EETQ"
+            )
+    elif quantize == "fp8":
+        from text_generation_server.layers.fp8 import Fp8Linear
+
+        linear = Fp8Linear(weight, bias)
+    elif quantize == "bitsandbytes":
+        try:
+            from text_generation_server.layers.bnb import (
+                warn_deprecate_bnb,
+                Linear8bitLt,
+            )
+        except ImportError:
+            raise NotImplementedError(
+                f"Bitsandbytes is missing install it with `pip install bitsandbytes`."
+            )
+        warn_deprecate_bnb()
+        linear = Linear8bitLt(
+            weight,
+            bias,
+            has_fp16_weights=False,
+            threshold=6.0,
+        )
+        if bias is not None:
+            linear.bias = nn.Parameter(bias)
+    elif quantize == "bitsandbytes-fp4":
+        try:
+            from text_generation_server.layers.bnb import Linear4bit
+        except ImportError:
+            raise NotImplementedError(
+                f"Bitsandbytes is missing install it with `pip install bitsandbytes`."
+            )
+        linear = Linear4bit(
+            weight,
+            bias,
+            quant_type="fp4",
+        )
+    elif quantize == "bitsandbytes-nf4":
+        try:
+            from text_generation_server.layers.bnb import Linear4bit
+        except ImportError:
+            raise NotImplementedError(
+                f"Bitsandbytes is missing install it with `pip install bitsandbytes`."
+            )
+        linear = Linear4bit(
+            weight,
+            bias,
+            quant_type="nf4",
+        )
+    elif quantize == "gptq":
+        try:
+            qweight, qzeros, scales, g_idx, bits, groupsize, use_exllama = weight
+        except Exception:
+            raise NotImplementedError(
+                f"The passed weight is not `gptq` compatible, loader needs to be updated."
+            )
+
+        if use_exllama:
+            try:
+                from text_generation_server.layers.gptq import (
+                    ExllamaQuantLinear,
+                )
+            except ImportError:
+                raise NotImplementedError(
+                    f"Exllama gptq kernels are not installed. Install them `cd server/exllama_kernels && python setup.py install && cd ../exllamav2_kernels && python setup.py install`"
+                )
+
+            linear = ExllamaQuantLinear(
+                qweight, qzeros, scales, g_idx, bias, bits, groupsize
+            )
+        else:
+            from text_generation_server.layers.gptq.quant_linear import QuantLinear
+
+            linear = QuantLinear(
+                qweight,
+                qzeros,
+                scales,
+                g_idx,
+                bias,
+                bits,
+                groupsize,
+            )
+    elif quantize == "awq":
+        try:
+            qweight, qzeros, scales, _, bits, groupsize, _ = weight
+        except Exception:
+            raise NotImplementedError(
+                f"The passed weight is not `awq` compatible, loader needs to be updated."
+            )
+        if SYSTEM == "rocm":
+            raise NotImplementedError(
+                "AWQ GEMM kernel can't be used on ROCm systems, please use `--quantize gptq` instead "
+                "to use Exllama/GPTQ kernels for AWQ inference."
+            )
+        try:
+            from text_generation_server.layers.awq.quantize.qmodule import WQLinear
+
+            linear = WQLinear(
+                w_bit=bits,
+                group_size=groupsize,
+                qweight=qweight,
+                qzeros=qzeros,
+                scales=scales,
+                bias=bias is not None,
+            )
+        except ImportError:
+            raise NotImplementedError(
+                "You do not seem to have awq installed, either install it (cd server &&  make install-awq), or try using GPTQ `---quantize gptq` a conversion AWQ->GPTQ will happen on the fly"
+            )
+    else:
+        raise NotImplementedError(f"Quantization `{quantize}` is not implemented yet.")
+    return linear

server/text_generation_server/layers/medusa.py

@@ -0,0 +1,186 @@
+import torch
+from torch import nn
+from typing import Tuple, Optional
+from text_generation_server.utils.speculate import get_speculate
+from text_generation_server.layers.linear import FastLinear
+from text_generation_server.layers.tensor_parallel import (
+    TensorParallelHead,
+    TensorParallelColumnLinear,
+)
+
+
+class ResBlock(torch.nn.Module):
+    def __init__(self, config, prefix, weights):
+        super().__init__()
+        self.linear = FastLinear.load(
+            config, prefix=f"{prefix}.linear", weights=weights, bias=True
+        )
+        self.act = torch.nn.SiLU()
+
+    def forward(self, x):
+        return x + self.act(self.linear(x))
+
+
+class MedusaModel(torch.nn.Module):
+    def __init__(self, config, medusa_config, weights):
+        super().__init__()
+        self.heads = torch.nn.ModuleList(
+            [
+                MedusaHead(config, medusa_config, prefix=f"{i}", weights=weights)
+                for i in range(get_speculate())
+            ]
+        )
+
+    def forward(self, x):
+        speculative_logits = torch.stack([head(x) for head in self.heads], dim=1)
+        return speculative_logits
+
+
+class MedusaHead(torch.nn.Module):
+    def __init__(self, config, medusa_config, prefix, weights):
+        super().__init__()
+        self.blocks = torch.nn.ModuleList(
+            [
+                ResBlock(config, prefix=f"{prefix}.{i}", weights=weights)
+                for i in range(medusa_config["medusa_num_layers"])
+            ]
+        )
+        n = len(self.blocks)
+        self.out = FastLinear.load(
+            config, prefix=f"{prefix}.{n}", weights=weights, bias=False
+        )
+
+    def forward(self, x):
+        for block in self.blocks:
+            x = block(x)
+        x = self.out(x)
+        return x
+
+
+class MedusaHeadV1(nn.Module):
+    def __init__(self, lm_head, medusa):
+        super().__init__()
+        self.lm_head = lm_head
+        self.medusa = medusa
+
+    @staticmethod
+    def load(config, prefix: str, weights):
+        from pathlib import Path
+        from safetensors import safe_open
+        import json
+
+        use_medusa = config.use_medusa
+
+        medusa_config = str(Path(use_medusa) / "config.json")
+        filename = str(Path(use_medusa) / "medusa_lm_head.safetensors")
+
+        with open(medusa_config, "r") as f:
+            medusa_config = json.load(f)
+        routing = weights.routing
+        with safe_open(filename, framework="pytorch") as f:
+            for k in f.keys():
+                if k in routing and routing[k] != filename:
+                    raise RuntimeError(
+                        f"Key {k} was found in multiple files: {filename} and {routing[k]}"
+                    )
+                routing[k] = filename
+
+        medusa = MedusaModel(config, medusa_config, weights)
+        lm_head = TensorParallelHead.load(config, prefix, weights)
+        return MedusaHeadV1(lm_head, medusa)
+
+    def forward(
+        self, input: torch.Tensor
+    ) -> Tuple[torch.Tensor, Optional[torch.Tensor]]:
+        logits = self.lm_head(input)
+        # If we have too many tokens, we skip speculative logits
+        if input.shape[0] > 128:
+            return logits, None
+
+        speculative_logits = self.medusa(input)
+        return logits, speculative_logits
+
+
+class MedusaHeadV2(nn.Module):
+    def __init__(self, config, prefix, weights):
+        super().__init__()
+        from pathlib import Path
+        from safetensors import safe_open
+        import json
+
+        use_medusa = config.use_medusa
+
+        medusa_config = str(Path(use_medusa) / "config.json")
+        filename = str(Path(use_medusa) / "medusa_lm_head.safetensors")
+
+        with open(medusa_config, "r") as f:
+            medusa_config = json.load(f)
+        routing = weights.routing
+        with safe_open(filename, framework="pytorch") as f:
+            for k in f.keys():
+                if k in routing and routing[k] != filename:
+                    raise RuntimeError(
+                        f"Key {k} was found in multiple files: {filename} and {routing[k]}"
+                    )
+                routing[k] = filename
+
+        self.n_medusa_heads = get_speculate()
+
+        assert medusa_config["medusa_num_layers"] == 1
+        self.linear = TensorParallelColumnLinear.load_multi(
+            config,
+            prefixes=[f"{i}.0.linear" for i in range(self.n_medusa_heads)],
+            dim=0,
+            weights=weights,
+            bias=True,
+        )
+        self.process_group = weights.process_group
+        self.world_size = self.process_group.size()
+        self.rank = self.process_group.rank()
+
+        self.act = torch.nn.SiLU()
+
+        self.lm_head = TensorParallelHead.load(config, prefix, weights)
+
+    def forward(self, x):
+        # If we have too many tokens, we skip speculative logits
+        if x.shape[0] > 128:
+            logits = self.lm_head(x)
+            return logits, None
+
+        size = x.shape[-1]
+        block_size = (size + self.world_size - 1) // self.world_size
+        start = self.rank * block_size
+        stop = (self.rank + 1) * block_size
+
+        x_block = x[:, start:stop]
+
+        # Compute all medusa heads at the same time, then reshape and move the n_medusa_heads dim to dim 1
+        medusa_res = self.act(self.linear(x)).reshape(
+            *x_block.shape[:-1], self.n_medusa_heads, x_block.shape[-1]
+        )
+
+        # Apply all residual medusa heads
+        output = x[:, start:stop].unsqueeze(-2) + medusa_res
+
+        # Gather medusa heads
+        world_output = [
+            torch.empty_like(output) for _ in range(self.process_group.size())
+        ]
+        torch.distributed.all_gather(world_output, output, group=self.process_group)
+        world_output = torch.cat(world_output, dim=-1)
+
+        # Stack x and medusa residual x
+        stacked_x = torch.cat([x.unsqueeze(-2), world_output], dim=-2)
+
+        # Compute lm head on x + medusa residual x
+        logits = self.lm_head(stacked_x)
+
+        # Finally, split logits from speculative logits
+        logits, speculative_logits = torch.split(
+            logits, [1, self.n_medusa_heads], dim=-2
+        )
+        # Squeeze added dimension
+        logits = logits.squeeze(-2)
+
+        return logits, speculative_logits

server/text_generation_server/layers/rotary.py

@@ -0,0 +1,419 @@
+import os
+import torch
+from torch import nn
+
+from text_generation_server.utils.import_utils import SYSTEM
+
+if SYSTEM == "cuda":
+    from flash_attn.layers.rotary import RotaryEmbedding
+    import rotary_emb
+elif SYSTEM == "rocm":
+    from vllm import pos_encoding_ops
+
+
+def _create_inv_freq(dim, base, device):
+    inv_freq = 1.0 / (
+        base ** (torch.arange(0, dim, 2, device=device, dtype=torch.float32) / dim)
+    )
+    return inv_freq
+
+
+def _get_rope_config(config):
+    if os.getenv("ROPE_SCALING", None) is not None:
+        rope_scaling = {
+            "type": os.environ["ROPE_SCALING"],
+            "factor": float(os.environ["ROPE_FACTOR"]),
+        }
+        return rope_scaling
+    return getattr(config, "rope_scaling", None)
+
+
+class PositionRotaryEmbedding(nn.Module):
+    def __init__(self, inv_freq, scaling_factor):
+        super().__init__()
+        self.inv_freq = inv_freq
+        self._seq_len_cached = 0
+        self._cos_cached = None
+        self._sin_cached = None
+        self._cos_k_cached = None
+        self._sin_k_cached = None
+        self.scaling_factor = scaling_factor
+        self.dynamic_args = None
+
+    def forward(
+        self,
+        query: torch.Tensor,
+        key: torch.Tensor,
+        cos: torch.Tensor,
+        sin: torch.Tensor,
+    ):
+        # Such controlflows may add some overhead.
+        if SYSTEM == "cuda":
+            rotary_dim = cos.shape[-1]
+            q1 = query[..., :rotary_dim]
+            q2 = query[..., rotary_dim : 2 * rotary_dim]
+
+            rotary_emb.apply_rotary(q1, q2, cos, sin, q1, q2, False)
+
+            k1 = key[..., :rotary_dim]
+            k2 = key[..., rotary_dim : 2 * rotary_dim]
+
+            rotary_emb.apply_rotary(k1, k2, cos, sin, k1, k2, False)
+        elif SYSTEM == "rocm":
+            # NOTE: On RoCm systems, we use a ROPE implementatation adapted from VLLM which launches a single kernel for both query/key, contrary to flash-attn implementation used on NVIDIA systems.
+            # Compiling flash-attn rotary on RoCm, it appears hipcc is unable to unroll loops, resulting in an even slower inference compared to eager: https://github.com/pytorch/pytorch/issues/113773
+
+            head_size = query.shape[-1]
+
+            # Inplace operation, updating query and key.
+            pos_encoding_ops.rotary_embedding(query, key, head_size, cos, sin, True)
+        elif SYSTEM == "xpu":
+            ipex.llm.functional.rotary_embedding(
+                query, key, sin, cos, query.size(-1), True
+            )
+        else:
+            raise ValueError(
+                "Your system seem to be not supported. Please check your install or open an issue at https://github.com/huggingface/text-generation-inference/issues with a clear reproduction."
+            )
+
+    @classmethod
+    def static(cls, config, dim, base, device):
+        inv_freq = _create_inv_freq(dim, base, device)
+        scaling_factor = None
+        rope_scaling = _get_rope_config(config)
+        if rope_scaling is not None:
+            if rope_scaling["type"] == "linear":
+                pass
+            elif rope_scaling["type"] == "dynamic":
+                scaling_factor = rope_scaling["factor"]
+                return DynamicPositionRotaryEmbedding(
+                    dim=dim,
+                    max_position_embeddings=config.max_position_embeddings,
+                    base=base,
+                    device=inv_freq.device,
+                    scaling_factor=scaling_factor,
+                )
+            elif rope_scaling["type"] == "yarn":
+                scaling_factor = rope_scaling["factor"]
+                return YarnPositionRotaryEmbedding(
+                    dim=2 * inv_freq.shape[0],
+                    max_position_embeddings=rope_scaling[
+                        "original_max_position_embeddings"
+                    ],
+                    base=10000.0,
+                    device=inv_freq.device,
+                    scaling_factor=scaling_factor,
+                    extrapolation_factor=1,
+                    attn_factor=1,
+                    beta_fast=32,
+                    beta_slow=1,
+                )
+            elif rope_scaling["type"] == "su":
+                short_factor = torch.tensor(
+                    rope_scaling["short_factor"], dtype=torch.float32, device=device
+                )
+                short_inv_freq = 1.0 / (
+                    short_factor
+                    * base
+                    ** (
+                        torch.arange(0, dim, 2, device=device, dtype=torch.float32)
+                        / dim
+                    )
+                )
+                long_factor = torch.tensor(
+                    rope_scaling["long_factor"], dtype=torch.float32, device=device
+                )
+                long_inv_freq = 1.0 / (
+                    long_factor
+                    * base
+                    ** (
+                        torch.arange(0, dim, 2, device=device, dtype=torch.float32)
+                        / dim
+                    )
+                )
+
+                original_max_position_embeddings = (
+                    config.original_max_position_embeddings
+                )
+                max_position_embeddings = config.max_position_embeddings
+                if max_position_embeddings <= original_max_position_embeddings:
+                    scaling_factor = 1.0
+                else:
+                    scale = max_position_embeddings / original_max_position_embeddings
+                    scaling_factor = math.sqrt(
+                        1 + math.log(scale) / math.log(original_max_position_embeddings)
+                    )
+
+                return SuRotaryEmbedding(
+                    short_inv_freq=short_inv_freq,
+                    long_inv_freq=long_inv_freq,
+                    scaling_factor=scaling_factor,
+                    original_max_position_embeddings=original_max_position_embeddings,
+                )
+            else:
+                raise NotImplementedError(
+                    f"rope scaling type {rope_scaling['type']} is not implemented or invalid"
+                )
+        return cls(inv_freq, scaling_factor)
+
+    @classmethod
+    def load(cls, config, prefix, weights):
+        # XXX: Always load this in float32 !
+        dtype = weights.dtype
+        weights.dtype = torch.float32
+        inv_freq = weights.get_tensor(f"{prefix}.inv_freq")
+        weights.dtype = dtype
+
+        scaling_factor = None
+        rope_scaling = _get_rope_config(config)
+        if rope_scaling is not None:
+            scaling_factor = rope_scaling["factor"]
+            if rope_scaling["type"] == "linear":
+                pass
+            elif rope_scaling["type"] == "dynamic":
+                return DynamicPositionRotaryEmbedding(
+                    dim=2 * inv_freq.shape[0],
+                    max_position_embeddings=config.max_position_embeddings,
+                    base=10000.0,
+                    device=inv_freq.device,
+                    scaling_factor=scaling_factor,
+                )
+            elif rope_scaling["type"] == "yarn":
+                return YarnPositionRotaryEmbedding(
+                    dim=2 * inv_freq.shape[0],
+                    max_position_embeddings=rope_scaling[
+                        "original_max_position_embeddings"
+                    ],
+                    base=10000.0,
+                    device=inv_freq.device,
+                    scaling_factor=scaling_factor,
+                    extrapolation_factor=1,
+                    attn_factor=1,
+                    beta_fast=32,
+                    beta_slow=1,
+                )
+            else:
+                raise NotImplementedError(
+                    f"rope scaling type {rope_scaling['type']} is not implemented or invalid"
+                )
+        return cls(inv_freq, scaling_factor)
+
+    def _update_cos_sin_cache(self, dtype, device, seqlen):
+        # Reset the tables if the sequence length has changed,
+        # or if we're on a new device (possibly due to tracing for instance)
+        if (
+            seqlen > self._seq_len_cached
+            or self._cos_cached.device != device
+            or self._cos_cached.dtype != dtype
+        ):
+            self._seq_len_cached = seqlen
+            t = torch.arange(seqlen, device=device, dtype=self.inv_freq.dtype)
+            if self.scaling_factor is not None:
+                t /= self.scaling_factor
+            # Don't do einsum, it converts fp32 to fp16
+            # freqs = torch.einsum("i,j->ij", t, self.inv_freq)
+
+            freqs = torch.outer(t, self.inv_freq.to(device=t.device))
+            self._cos_cached = torch.cos(freqs).to(dtype)
+            self._sin_cached = torch.sin(freqs).to(dtype)
+
+    def get_cos_sin(self, position_ids: torch.Tensor, max_s: int, dtype: torch.dtype):
+        """
+        Return cos and sin for the asked position ids
+        """
+        if SYSTEM == "rocm":
+            # For RoCm, we always use float cos/sin to avoid a cast.
+            # For NVIDIA, for some reason, the flash-attn rotary kernel requires cos/sin and query/key to be of same dtype: https://github.com/Dao-AILab/flash-attention/blob/017716451d446e464dde9aca3a3c1ed2209caaa9/csrc/rotary/rotary.cpp#L26
+            # But later on goes and cast cos/sin to float anyway: https://github.com/Dao-AILab/flash-attention/blob/017716451d446e464dde9aca3a3c1ed2209caaa9/csrc/rotary/rotary_cuda.cu#L29, which looks suboptimal.
+            dtype = torch.float32
+
+        self._update_cos_sin_cache(dtype, position_ids.device, max_s)
+
+        cos = torch.index_select(self._cos_cached, 0, position_ids)
+        sin = torch.index_select(self._sin_cached, 0, position_ids)
+
+        # Note: this unsqueeze is not necessary on RoCm + VLLM ROPE implementation, but we leave it as is to avoid yet an other controlflow.
+        return cos.unsqueeze(1), sin.unsqueeze(1)
+
+
+class SuRotaryEmbedding(PositionRotaryEmbedding):
+    def __init__(
+        self,
+        short_inv_freq,
+        long_inv_freq,
+        scaling_factor,
+        original_max_position_embeddings,
+    ):
+        super(PositionRotaryEmbedding, self).__init__()
+        self.short_inv_freq = short_inv_freq
+        self.long_inv_freq = long_inv_freq
+        self.scaling_factor = scaling_factor
+        self.original_max_position_embeddings = original_max_position_embeddings
+        self._seq_len_cached = 0
+        self._cos_cached = None
+        self._sin_cached = None
+        self._cos_k_cached = None
+        self._sin_k_cached = None
+        self.dynamic_args = None
+
+    def _update_cos_sin_cache(self, dtype, device, seqlen):
+        # Reset the tables if the sequence length has changed,
+        # or if we're on a new device (possibly due to tracing for instance)
+        if (
+            seqlen > self._seq_len_cached
+            or self._cos_cached.device != device
+            or self._cos_cached.dtype != dtype
+        ):
+            self._seq_len_cached = seqlen
+            if seqlen > self.original_max_position_embeddings:
+                inv_freq = self.long_inv_freq
+            else:
+                inv_freq = self.short_inv_freq
+            t = torch.arange(seqlen, device=device, dtype=inv_freq.dtype)
+            if self.scaling_factor is not None:
+                t /= self.scaling_factor
+            # Don't do einsum, it converts fp32 to fp16
+            # freqs = torch.einsum("i,j->ij", t, self.inv_freq)
+
+            freqs = torch.outer(t, inv_freq.to(device=t.device))
+            self._cos_cached = torch.cos(freqs).to(dtype)
+            self._sin_cached = torch.sin(freqs).to(dtype)
+
+
+class DynamicPositionRotaryEmbedding(PositionRotaryEmbedding):
+    def __init__(self, dim, max_position_embeddings, base, device, scaling_factor):
+        inv_freq = _create_inv_freq(dim, base, device)
+        super().__init__(inv_freq, scaling_factor)
+        self.dim = dim
+        self.max_position_embeddings = max_position_embeddings
+        self.base = base
+
+    def _update_cos_sin_cache(self, dtype, device, seqlen):
+        # Reset the tables if the sequence length has changed,
+        # or if we're on a new device (possibly due to tracing for instance)
+        if (
+            seqlen > self._seq_len_cached
+            or self._cos_cached.device != device
+            or self._cos_cached.dtype != dtype
+        ):
+            if seqlen > self.max_position_embeddings:
+                newbase = self.base * (
+                    (self.scaling_factor * seqlen / self.max_position_embeddings)
+                    - (self.scaling_factor - 1)
+                ) ** (self.dim / (self.dim - 2))
+                self.inv_freq = _create_inv_freq(
+                    self.dim, newbase, self.inv_freq.device
+                )
+            self._seq_len_cached = seqlen
+            t = torch.arange(seqlen, device=device, dtype=self.inv_freq.dtype)
+            # Don't do einsum, it converts fp32 to fp16
+            # freqs = torch.einsum("i,j->ij", t, self.inv_freq)
+
+            freqs = torch.outer(t, self.inv_freq.to(device=t.device))
+            self._cos_cached = torch.cos(freqs).to(dtype)
+            self._sin_cached = torch.sin(freqs).to(dtype)
+
+
+# Inverse dim formula to find dim based on number of rotations
+import math
+
+
+def find_correction_dim(num_rotations, dim, base=10000, max_position_embeddings=2048):
+    return (dim * math.log(max_position_embeddings / (num_rotations * 2 * math.pi))) / (
+        2 * math.log(base)
+    )
+
+
+# Find dim range bounds based on rotations
+def find_correction_range(
+    low_rot, high_rot, dim, base=10000, max_position_embeddings=2048
+):
+    low = math.floor(find_correction_dim(low_rot, dim, base, max_position_embeddings))
+    high = math.ceil(find_correction_dim(high_rot, dim, base, max_position_embeddings))
+    return max(low, 0), min(high, dim - 1)  # Clamp values just in case
+
+
+def linear_ramp_mask(min, max, dim):
+    if min == max:
+        max += 0.001  # Prevent singularity
+
+    linear_func = (torch.arange(dim, dtype=torch.float32) - min) / (max - min)
+    ramp_func = torch.clamp(linear_func, 0, 1)
+    return ramp_func
+
+
+def get_mscale(scale=1):
+    if scale <= 1:
+        return 1.0
+    return 0.1 * math.log(scale) + 1.0
+
+
+class YarnPositionRotaryEmbedding(PositionRotaryEmbedding):
+    def __init__(
+        self,
+        dim,
+        max_position_embeddings,
+        base,
+        device,
+        scaling_factor,
+        *,
+        extrapolation_factor,
+        attn_factor,
+        beta_fast,
+        beta_slow,
+    ):
+        inv_freq = _create_inv_freq(dim, base, device)
+        super().__init__(inv_freq, scaling_factor)
+        self.dim = dim
+        self.max_position_embeddings = max_position_embeddings
+        self.base = base
+        self.extrapolation_factor = extrapolation_factor
+        self.attn_factor = attn_factor
+        self.beta_fast = beta_fast
+        self.beta_slow = beta_slow
+        self.mscale = float(
+            get_mscale(self.scaling_factor) * self.attn_factor
+        )  # Get n-d magnitude scaling corrected for interpolation
+
+    def _update_cos_sin_cache(self, dtype, device, seqlen):
+        # Reset the tables if the sequence length has changed,
+        # or if we're on a new device (possibly due to tracing for instance)
+        if (
+            seqlen > self._seq_len_cached
+            or self._cos_cached.device != device
+            or self._cos_cached.dtype != dtype
+        ):
+            if seqlen > self.max_position_embeddings:
+                inv_freq_extrapolation = _create_inv_freq(
+                    self.dim, self.base, self.inv_freq.device
+                )
+                freqs = 1.0 / inv_freq_extrapolation
+                inv_freq_interpolation = 1.0 / (self.scaling_factor * freqs)
+                low, high = find_correction_range(
+                    self.beta_fast,
+                    self.beta_slow,
+                    self.dim,
+                    self.base,
+                    self.max_position_embeddings,
+                )
+                inv_freq_mask = (
+                    1 - linear_ramp_mask(low, high, self.dim // 2).float().to(device)
+                ) * self.extrapolation_factor  # Get n-d rotational scaling corrected for extrapolation
+                inv_freq = (
+                    inv_freq_interpolation * (1 - inv_freq_mask)
+                    + inv_freq_extrapolation * inv_freq_mask
+                )
+
+                self.inv_freq = inv_freq
+                self.mscale = float(
+                    get_mscale(self.scaling_factor) * self.attn_factor
+                )  # Get n-d magnitude scaling corrected for interpolation
+
+            self._seq_len_cached = seqlen
+            t = torch.arange(seqlen, device=device, dtype=self.inv_freq.dtype)
+            # Don't do einsum, it converts fp32 to fp16
+            # freqs = torch.einsum("i,j->ij", t, self.inv_freq)
+
+            freqs = torch.outer(t, self.inv_freq.to(device=t.device))
+            self._cos_cached = (torch.cos(freqs) * self.mscale).to(dtype)
+            self._sin_cached = (torch.sin(freqs) * self.mscale).to(dtype)

server/text_generation_server/layers/speculative.py

@@ -0,0 +1,35 @@
+import torch
+from typing import Tuple, Optional
+from text_generation_server.layers.medusa import MedusaHeadV1, MedusaHeadV2
+from text_generation_server.layers.tensor_parallel import TensorParallelHead
+
+
+class SpeculativeHead(torch.nn.Module):
+    def __init__(self, lm_head, medusa):
+        super().__init__()
+        self.head = lm_head
+        self.medusa = medusa
+
+    @staticmethod
+    def load(config, prefix: str, weights):
+        use_medusa = config.use_medusa
+        if use_medusa:
+            lm_head = None
+            try:
+                medusa = MedusaHeadV1.load(config, prefix, weights)
+            except:
+                medusa = MedusaHeadV2(config, prefix, weights)
+        else:
+            lm_head = TensorParallelHead.load(config, prefix, weights)
+            medusa = None
+        return SpeculativeHead(lm_head, medusa)
+
+    def forward(
+        self, input: torch.Tensor
+    ) -> Tuple[torch.Tensor, Optional[torch.Tensor]]:
+        if self.medusa is not None:
+            return self.medusa(input)
+
+        assert self.head is not None
+        logits = self.head(input)
+        return logits, None

server/text_generation_server/layers/tensor_parallel.py

@@ -0,0 +1,188 @@
+import torch
+from torch.nn import functional as F
+from typing import List
+from text_generation_server.layers.linear import get_linear, FastLinear
+
+
+class SuperLayer(torch.nn.Module):
+    def __init__(self, linear):
+        super().__init__()
+        self.linear = linear
+
+    def forward(self, x):
+        return self.linear.forward(x)
+
+
+class TensorParallelHead(SuperLayer):
+    def __init__(self, linear, process_group, should_gather: bool):
+        super().__init__(linear)
+        self.process_group = process_group
+        self.should_gather = should_gather
+
+    @staticmethod
+    def load(config, prefix: str, weights):
+        if weights.process_group.size() > 1:
+            try:
+                weight = weights.get_sharded(f"{prefix}.weight", dim=0)
+                should_gather = True
+            except AssertionError:
+                # If the vocab size is not divisible by number of shards
+                # just load the entire thing.
+                weight = weights.get_tensor(f"{prefix}.weight")
+                should_gather = False
+        else:
+            weight = weights.get_tensor(f"{prefix}.weight")
+            should_gather = False
+
+        # GPTQ,AWQ,EETQ don't quantize heads (nor embeddings)
+        if config.quantize in ["gptq", "awq", "eetq"]:
+            quantize = None
+        else:
+            quantize = config.quantize
+        return TensorParallelHead(
+            get_linear(weight, bias=None, quantize=quantize),
+            process_group=weights.process_group,
+            should_gather=should_gather,
+        )
+
+    def forward(self, input: torch.Tensor) -> torch.Tensor:
+        if not self.should_gather:
+            return super().forward(input)
+
+        world_size = self.process_group.size()
+        if len(input.shape) == 2 and isinstance(self.linear, FastLinear):
+            out_dim = self.linear.weight.shape[0]
+
+            if input.shape[0] == 1:
+                world_out = input.new_empty(1, out_dim * world_size)
+                local_out = input.new_empty(1, out_dim)
+                gather_input = local_out
+            else:
+                world_out = input.new_empty(out_dim * world_size, input.shape[0])
+                gather_input = input.new_empty(out_dim, input.shape[0])
+                local_out = gather_input.T
+
+            torch.mm(input, self.linear.weight.T, out=local_out)
+
+            torch.distributed.all_gather_into_tensor(
+                world_out, gather_input, group=self.process_group
+            )
+
+            if input.shape[0] == 1:
+                return world_out
+            return world_out.T
+
+        output = super().forward(input)
+        world_output = [
+            torch.empty_like(output) for _ in range(self.process_group.size())
+        ]
+        torch.distributed.all_gather(world_output, output, group=self.process_group)
+        world_output = torch.cat(world_output, dim=-1)
+        return world_output
+
+
+class TensorParallelColumnLinear(SuperLayer):
+    @classmethod
+    def load_gate_up(cls, config, prefix: str, weights, bias: bool):
+        """Specific method when the QKV was joined after the fact"""
+        weight = weights.get_weights_col_packed_gate_up(
+            prefix, quantize=config.quantize
+        )
+        if bias:
+            raise NotImplementedError("packed_gate_up only implemented without bias")
+        else:
+            bias = None
+        linear = get_linear(weight, bias, config.quantize)
+        return cls(linear)
+
+    @classmethod
+    def load_qkv(cls, config, prefix: str, weights, bias: bool):
+        """Specific method when the QKV was joined after the fact"""
+        weight = weights.get_weights_col_packed_qkv(prefix, quantize=config.quantize)
+        if bias:
+            raise NotImplementedError("packed_qkv only implemented for baichuan")
+        else:
+            bias = None
+        linear = get_linear(weight, bias, config.quantize)
+        return cls(linear)
+
+    @classmethod
+    def load(cls, config, prefix: str, weights, bias: bool):
+        return cls.load_multi(config, [prefix], weights, bias, dim=0)
+
+    @classmethod
+    def load_multi(cls, config, prefixes: List[str], weights, bias: bool, dim: int):
+        weight = weights.get_multi_weights_col(
+            prefixes, quantize=config.quantize, dim=dim
+        )
+
+        if bias:
+            b = [weights.get_sharded(f"{p}.bias", dim=0) for p in prefixes]
+            bias = torch.cat(b, dim=dim)
+        else:
+            bias = None
+        linear = get_linear(weight, bias, config.quantize)
+        return cls(linear)
+
+
+class TensorParallelRowLinear(SuperLayer):
+    def __init__(self, linear, process_group):
+        super().__init__(linear)
+        self.process_group = process_group
+
+    @classmethod
+    def load(cls, config, prefix: str, weights, bias: bool):
+        weight = weights.get_multi_weights_row(prefix, quantize=config.quantize)
+
+        if bias and weights.process_group.rank() == 0:
+            # Rank is only on the first rank process
+            bias = weights.get_tensor(f"{prefix}.bias")
+        else:
+            bias = None
+        return cls(
+            get_linear(weight, bias, config.quantize),
+            process_group=weights.process_group,
+        )
+
+    def forward(self, input: torch.Tensor, reduce: bool = True) -> torch.Tensor:
+        out = super().forward(input)
+        if self.process_group.size() > 1 and reduce:
+            torch.distributed.all_reduce(out, group=self.process_group)
+        return out
+
+
+class TensorParallelEmbedding(torch.nn.Module):
+    def __init__(self, prefix: str, weights, reduce=True):
+        super().__init__()
+        weight = weights.get_partial_sharded(f"{prefix}.weight", dim=0)
+        num_embeddings = weights.get_shape(f"{prefix}.weight")[0]
+
+        process_group = weights.process_group
+
+        world_size = process_group.size()
+        rank = process_group.rank()
+
+        block_size = (num_embeddings + world_size - 1) // world_size
+        self.min_id = rank * block_size
+        self.max_id = min(num_embeddings, (rank + 1) * block_size)
+        self.null_idx = weight.shape[
+            0
+        ]  # Usually block_size, might be less in non even vocab_size.
+        self.process_group = weights.process_group
+        self.reduce = reduce
+
+        """Additional 0 entry used for masking"""
+        self.weight = torch.nn.Parameter(F.pad(weight, (0, 0, 0, 1)))
+
+    def forward(self, input: torch.Tensor) -> torch.Tensor:
+        # default all out of bounds values to `self.null_idx` that will then be mapped to 0
+        # translate for [0, self.max_id - self.min_id[
+        input = torch.where(
+            (self.min_id > input) | (input >= self.max_id),
+            self.null_idx,
+            input - self.min_id,
+        )
+        out = torch.nn.functional.embedding(input, self.weight)
+        if self.reduce and self.process_group.size() > 1:
+            torch.distributed.all_reduce(out, group=self.process_group)
+        return out

server/text_generation_server/models/cache_manager.py

@@ -2,7 +2,7 @@ import math
 import torch
 
 from typing import Optional, List, Tuple
-from text_generation_server.utils.import_utils import IS_XPU_SYSTEM
+from text_generation_server.utils.import_utils import SYSTEM
 
 BLOCK_SIZE: int = 16
 # Will be set in warmup
@@ -25,7 +25,7 @@ class CacheManager:
         self.repeat_slots = repeat_slots
 
         element_size = torch.tensor([], dtype=dtype).element_size()
-        if IS_XPU_SYSTEM:
+        if SYSTEM == "xpu":
             x = 1
         else:
             x = self.block_size // element_size

server/text_generation_server/models/custom_modeling/bloom_modeling.py

@@ -32,7 +32,7 @@ from transformers.modeling_outputs import (
 )
 from transformers import BloomConfig, PreTrainedModel
 
-from text_generation_server.utils.layers import (
+from text_generation_server.layers import (
     TensorParallelColumnLinear,
     TensorParallelEmbedding,
     TensorParallelRowLinear,

server/text_generation_server/models/custom_modeling/clip.py

@@ -15,7 +15,7 @@ from transformers.modeling_outputs import (
 )
 from transformers import CLIPConfig, CLIPTextConfig, CLIPVisionConfig
 
-from text_generation_server.utils.layers import (
+from text_generation_server.layers import (
     TensorParallelEmbedding,
     TensorParallelColumnLinear,
     TensorParallelRowLinear,

server/text_generation_server/models/custom_modeling/flash_cohere_modeling.py

@@ -26,18 +26,22 @@ from transformers.activations import ACT2FN
 from typing import Optional, List, Tuple
 
 from text_generation_server.utils import paged_attention, flash_attn
-from text_generation_server.utils.import_utils import IS_ROCM_SYSTEM, IS_CUDA_SYSTEM
+from text_generation_server.utils.import_utils import SYSTEM
-from text_generation_server.utils.layers import (
+from text_generation_server.layers import (
     TensorParallelRowLinear,
     TensorParallelColumnLinear,
     TensorParallelEmbedding,
-    PositionRotaryEmbedding,
     SpeculativeHead,
     get_linear,
+)
+from text_generation_server.layers.layernorm import (
     FastLayerNorm,
 )
+from text_generation_server.layers.rotary import (
+    PositionRotaryEmbedding,
+)
 
-if IS_CUDA_SYSTEM:
+if SYSTEM == "cuda":
     import dropout_layer_norm
 else:
     dropout_layer_norm = None
@@ -52,7 +56,7 @@ class CohereRotary(PositionRotaryEmbedding):
         sin: torch.Tensor,
     ):
         # Such controlflows may add some overhead.
-        if IS_CUDA_SYSTEM:
+        if SYSTEM == "cuda":
             import rotary_emb
 
             q1 = query[..., ::2]
@@ -64,7 +68,7 @@ class CohereRotary(PositionRotaryEmbedding):
             k2 = key[..., 1::2]
 
             rotary_emb.apply_rotary(k1, k2, cos, sin, k1, k2, False)
-        elif IS_ROCM_SYSTEM:
+        elif SYSTEM == "rocm":
             from vllm import pos_encoding_ops
 
             # NOTE: On RoCm systems, we use a ROPE implementatation adapted from VLLM which launches a single kernel for both query/key, contrary to flash-attn implementation used on NVIDIA systems.
@@ -90,7 +94,7 @@ class CohereLayerNorm(nn.Module):
         self.eps = eps
 
     def forward(self, hidden_states):
-        if hidden_states.shape[-1] > 8192 or IS_ROCM_SYSTEM:
+        if hidden_states.shape[-1] > 8192 or SYSTEM == "rocm":
             hidden_states = hidden_states.reshape(
                 -1, self.weight.shape[0], self.weight.shape[1]
             )

server/text_generation_server/models/custom_modeling/flash_dbrx_modeling.py

@@ -21,21 +21,26 @@ from transformers.activations import ACT2FN
 from transformers.configuration_utils import PretrainedConfig
 from typing import Optional, List, Tuple, Any
 from loguru import logger
-from text_generation_server.utils.import_utils import IS_XPU_SYSTEM
+from text_generation_server.utils.import_utils import SYSTEM
 
-if not IS_XPU_SYSTEM:
+if SYSTEM != "xpu":
     from vllm.model_executor.layers.fused_moe import fused_moe
+
 from text_generation_server.utils import paged_attention, flash_attn
-from text_generation_server.utils.layers import (
+from text_generation_server.layers import (
     FastLinear,
-    FastLayerNorm,
     TensorParallelRowLinear,
     TensorParallelColumnLinear,
     TensorParallelEmbedding,
-    PositionRotaryEmbedding,
     SpeculativeHead,
     get_linear,
 )
+from text_generation_server.layers.rotary import (
+    PositionRotaryEmbedding,
+)
+from text_generation_server.layers.layernorm import (
+    FastLayerNorm,
+)
 from text_generation_server.utils.log import log_once
 
 
@@ -216,7 +221,7 @@ def _load_gqa(config, prefix: str, weights):
 
         bits, groupsize, desc_act, quant_method = weights._get_gptq_params()
 
-        from text_generation_server.utils.layers import HAS_EXLLAMA
+        from text_generation_server.layers import HAS_EXLLAMA
 
         use_exllama = (
             bits == 4 and HAS_EXLLAMA and config.quantize == "gptq" and not desc_act
@@ -236,7 +241,7 @@ def _load_gqa(config, prefix: str, weights):
             log_once(
                 logger.info, "Converting AWQ model to Exllama/GPTQ packing format."
             )
-            from text_generation_server.utils.awq.conversion_utils import (
+            from text_generation_server.layers.awq.conveersion_utils import (
                 fast_awq_to_gptq,
             )
 

server/text_generation_server/models/custom_modeling/flash_gemma_modeling.py

@@ -27,13 +27,15 @@ from transformers.configuration_utils import PretrainedConfig
 from typing import Optional, List, Tuple
 
 from text_generation_server.utils import paged_attention, flash_attn
-from text_generation_server.utils.layers import (
+from text_generation_server.layers import (
     TensorParallelRowLinear,
     TensorParallelColumnLinear,
     TensorParallelEmbedding,
-    PositionRotaryEmbedding,
     SpeculativeHead,
     get_linear,
+)
+from text_generation_server.layers.rotary import PositionRotaryEmbedding
+from text_generation_server.layers.layernorm import (
     FastRMSNorm,
 )
 

server/text_generation_server/models/custom_modeling/flash_llama_modeling.py

@@ -27,13 +27,15 @@ from transformers.configuration_utils import PretrainedConfig
 from typing import Optional, List, Tuple
 
 from text_generation_server.utils import paged_attention, flash_attn
-from text_generation_server.utils.layers import (
+from text_generation_server.layers import (
     TensorParallelRowLinear,
     TensorParallelColumnLinear,
     TensorParallelEmbedding,
-    PositionRotaryEmbedding,
     SpeculativeHead,
     get_linear,
+)
+from text_generation_server.layers.rotary import PositionRotaryEmbedding
+from text_generation_server.layers.layernorm import (
     FastRMSNorm,
 )
 

server/text_generation_server/models/custom_modeling/flash_mistral_modeling.py

@@ -27,13 +27,15 @@ from transformers.configuration_utils import PretrainedConfig
 from typing import Optional, List, Tuple
 
 from text_generation_server.utils import paged_attention, flash_attn
-from text_generation_server.utils.layers import (
+from text_generation_server.layers import (
     TensorParallelRowLinear,
     TensorParallelColumnLinear,
     TensorParallelEmbedding,
-    PositionRotaryEmbedding,
     SpeculativeHead,
     get_linear,
+)
+from text_generation_server.layers.rotary import PositionRotaryEmbedding
+from text_generation_server.layers.layernorm import (
     FastRMSNorm,
 )
 

server/text_generation_server/models/custom_modeling/flash_mixtral_modeling.py

@@ -24,9 +24,9 @@ import torch.distributed
 import numpy as np
 
 from torch import nn
-from text_generation_server.utils.import_utils import IS_XPU_SYSTEM
+from text_generation_server.utils.import_utils import SYSTEM
 
-if not IS_XPU_SYSTEM:
+if SYSTEM != "xpu":
     from vllm.model_executor.layers.fused_moe import fused_moe
 from transformers.activations import ACT2FN
 from transformers.configuration_utils import PretrainedConfig
@@ -34,16 +34,20 @@ from typing import Optional, List, Tuple
 from loguru import logger
 
 from text_generation_server.utils import paged_attention, flash_attn
-from text_generation_server.utils.layers import (
+from text_generation_server.layers import (
     FastLinear,
-    FastRMSNorm,
     TensorParallelRowLinear,
     TensorParallelColumnLinear,
     TensorParallelEmbedding,
-    PositionRotaryEmbedding,
     SpeculativeHead,
     get_linear,
 )
+from text_generation_server.layers.layernorm import (
+    FastRMSNorm,
+)
+from text_generation_server.layers.rotary import (
+    PositionRotaryEmbedding,
+)
 
 
 class MixtralConfig(PretrainedConfig):

server/text_generation_server/models/custom_modeling/flash_neox_modeling.py

@@ -29,15 +29,19 @@ from typing import Optional, List, Tuple
 
 from text_generation_server.utils import paged_attention, flash_attn
 from text_generation_server.utils.flash_attn import attention
-from text_generation_server.utils.layers import (
+from text_generation_server.layers import (
     TensorParallelRowLinear,
     TensorParallelColumnLinear,
     TensorParallelEmbedding,
     SpeculativeHead,
-    FastLayerNorm,
-    PositionRotaryEmbedding,
     get_linear,
 )
+from text_generation_server.layers.layernorm import (
+    FastLayerNorm,
+)
+from text_generation_server.layers.rotary import (
+    PositionRotaryEmbedding,
+)
 
 
 def load_row(config, prefix: str, weights, bias: bool):

server/text_generation_server/models/custom_modeling/flash_phi_modeling.py

@@ -7,15 +7,19 @@ from transformers.configuration_utils import PretrainedConfig
 from typing import Optional, List, Tuple
 
 from text_generation_server.utils import paged_attention, flash_attn
-from text_generation_server.utils.layers import (
+from text_generation_server.layers import (
     TensorParallelRowLinear,
     TensorParallelColumnLinear,
     TensorParallelEmbedding,
-    PositionRotaryEmbedding,
     SpeculativeHead,
     get_linear,
+)
+from text_generation_server.layers.layernorm import (
     FastLayerNorm,
 )
+from text_generation_server.layers.rotary import (
+    PositionRotaryEmbedding,
+)
 
 
 class PhiConfig(PretrainedConfig):

server/text_generation_server/models/custom_modeling/flash_qwen2_modeling.py

@@ -6,13 +6,15 @@ from transformers.activations import ACT2FN
 from typing import Optional, List, Tuple
 
 from text_generation_server.utils import paged_attention, flash_attn
-from text_generation_server.utils.layers import (
+from text_generation_server.layers import (
     TensorParallelRowLinear,
     TensorParallelColumnLinear,
     TensorParallelEmbedding,
-    PositionRotaryEmbedding,
     SpeculativeHead,
     get_linear,
+)
+from text_generation_server.layers.rotary import PositionRotaryEmbedding
+from text_generation_server.layers.layernorm import (
     FastRMSNorm,
 )
 

server/text_generation_server/models/custom_modeling/flash_rw_modeling.py

@@ -8,15 +8,19 @@ from typing import Optional, List, Tuple
 
 from text_generation_server.utils import paged_attention, flash_attn
 from text_generation_server.utils.flash_attn import attention
-from text_generation_server.utils.layers import (
+from text_generation_server.layers import (
     TensorParallelRowLinear,
     TensorParallelColumnLinear,
     TensorParallelEmbedding,
     SpeculativeHead,
-    FastLayerNorm,
-    PositionRotaryEmbedding,
     get_linear,
 )
+from text_generation_server.layers.layernorm import (
+    FastLayerNorm,
+)
+from text_generation_server.layers.rotary import (
+    PositionRotaryEmbedding,
+)
 
 
 def load_row(config, prefix: str, weights, bias: bool):

server/text_generation_server/models/custom_modeling/flash_santacoder_modeling.py

@@ -6,14 +6,16 @@ from transformers.activations import ACT2FN
 from typing import Optional, List, Tuple
 
 from text_generation_server.utils import paged_attention, flash_attn
-from text_generation_server.utils.layers import (
+from text_generation_server.layers import (
     TensorParallelRowLinear,
     TensorParallelColumnLinear,
     SpeculativeHead,
     TensorParallelEmbedding,
-    FastLayerNorm,
     get_linear,
 )
+from text_generation_server.layers.layernorm import (
+    FastLayerNorm,
+)
 
 
 def load_multi_mqa(
@@ -80,13 +82,13 @@ def _load_multi_mqa_gptq(
             g_idx = g_idx.to(device=weights.device)
         elif quant_method == "awq":
             g_idx = None
-            from text_generation_server.utils.awq.conversion_utils import (
+            from text_generation_server.layers.awq.conversion_utils import (
                 fast_awq_to_gptq,
             )
 
             qweight, qzeros = fast_awq_to_gptq(qweight, qzeros)
 
-        from text_generation_server.utils.layers import HAS_EXLLAMA
+        from text_generation_server.layers.gptq import HAS_EXLLAMA
 
         use_exllama = HAS_EXLLAMA
         weight = (qweight, qzeros, scales, g_idx, bits, groupsize, use_exllama)

server/text_generation_server/models/custom_modeling/flash_starcoder2_modeling.py

@@ -27,15 +27,19 @@ from transformers.configuration_utils import PretrainedConfig
 from typing import Optional, List, Tuple
 
 from text_generation_server.utils import paged_attention, flash_attn
-from text_generation_server.utils.layers import (
+from text_generation_server.layers import (
     TensorParallelRowLinear,
     TensorParallelColumnLinear,
     TensorParallelEmbedding,
-    PositionRotaryEmbedding,
     SpeculativeHead,
     get_linear,
-    FastRMSNorm,
+)
+from text_generation_server.layers.layernorm import (
     FastLayerNorm,
+    FastRMSNorm,
+)
+from text_generation_server.layers.rotary import (
+    PositionRotaryEmbedding,
 )
 
 

server/text_generation_server/models/custom_modeling/idefics2.py

@@ -29,7 +29,7 @@ from text_generation_server.models.custom_modeling.vlm import (
 )
 from transformers.modeling_attn_mask_utils import _prepare_4d_attention_mask
 
-from text_generation_server.utils.layers import (
+from text_generation_server.layers import (
     TensorParallelColumnLinear,
     TensorParallelEmbedding,
     TensorParallelRowLinear,

server/text_generation_server/models/custom_modeling/idefics_modeling.py

@@ -47,20 +47,22 @@ from text_generation_server.models.custom_modeling.idefics_vision import (
 from text_generation_server.models.custom_modeling.idefics_perceiver import (
     IdeficsPerceiverResampler,
 )
-from text_generation_server.utils.layers import (
+from text_generation_server.layers import (
     TensorParallelColumnLinear,
     TensorParallelEmbedding,
     TensorParallelRowLinear,
     SpeculativeHead,
-    PositionRotaryEmbedding,
     FastLinear,
 )
-from text_generation_server.utils.import_utils import IS_CUDA_SYSTEM, IS_ROCM_SYSTEM
+from text_generation_server.layers.rotary import PositionRotaryEmbedding
+from text_generation_server.utils.import_utils import SYSTEM
 
-if IS_CUDA_SYSTEM:
+if SYSTEM == "cuda":
     import dropout_layer_norm
-elif IS_ROCM_SYSTEM:
+elif SYSTEM == "rocm":
     from vllm import layernorm_ops
+else:
+    raise RuntimeError(f"Unsupported system {SYSTEM}")
 
 
 @dataclass
@@ -373,7 +375,7 @@ class IdeficsRMSNorm(nn.Module):
                 hidden_states = hidden_states.to(self.weight.dtype)
 
             return self.weight * hidden_states
-        elif IS_CUDA_SYSTEM:
+        elif SYSTEM == "cuda":
             # faster post attention rms norm
             unwrap = False
             if len(hidden_states.shape) > 2:
@@ -405,7 +407,7 @@ class IdeficsRMSNorm(nn.Module):
                 normed_hidden_states = normed_hidden_states.view(*shape)
 
             return normed_hidden_states
-        elif IS_ROCM_SYSTEM:
+        elif SYSTEM == "rocm":
             # We use VLLM RMSNorm kernel that can be compiled for RoCm, instead of Flash Attention ones that can not.
             if residual is not None:
                 hidden_states += residual

server/text_generation_server/models/custom_modeling/idefics_perceiver.py

@@ -41,7 +41,7 @@ from typing import Optional, Tuple
 import torch
 import torch.nn as nn
 
-from text_generation_server.utils.layers import (
+from text_generation_server.layers import (
     TensorParallelColumnLinear,
     TensorParallelRowLinear,
 )

server/text_generation_server/models/custom_modeling/idefics_vision.py

@@ -28,7 +28,7 @@ from transformers.utils import (
     ModelOutput,
     logging,
 )
-from text_generation_server.utils.layers import (
+from text_generation_server.layers import (
     TensorParallelColumnLinear,
     TensorParallelRowLinear,
     TensorParallelEmbedding,

server/text_generation_server/models/custom_modeling/llava_next.py

@@ -27,7 +27,7 @@ from text_generation_server.models.custom_modeling.vlm import (
     load_text_model,
     load_vision_model,
 )
-from text_generation_server.utils.layers import (
+from text_generation_server.layers import (
     TensorParallelColumnLinear,
     TensorParallelRowLinear,
 )

server/text_generation_server/models/custom_modeling/mamba_modeling.py

@@ -8,12 +8,12 @@ from typing import Optional, Tuple, Any
 from transformers.configuration_utils import PretrainedConfig
 import torch.nn.functional as F
 
-from text_generation_server.utils.layers import (
+from text_generation_server.layers import (
     SpeculativeHead,
     TensorParallelEmbedding,
-    FastRMSNorm,
     FastLinear,
 )
+from text_generation_server.layers.layernorm import FastRMSNorm
 
 from einops import rearrange
 from causal_conv1d import causal_conv1d_fn, causal_conv1d_update

server/text_generation_server/models/custom_modeling/mpt_modeling.py

@@ -17,7 +17,7 @@ from transformers.modeling_outputs import (
 )
 from einops import rearrange
 from packaging import version
-from text_generation_server.utils.layers import (
+from text_generation_server.layers import (
     TensorParallelEmbedding,
     TensorParallelColumnLinear,
     TensorParallelRowLinear,

server/text_generation_server/models/custom_modeling/neox_modeling.py

@@ -40,7 +40,7 @@ from transformers.modeling_outputs import (
 from transformers.modeling_utils import PreTrainedModel
 from transformers import GPTNeoXConfig
 from loguru import logger
-from text_generation_server.utils.layers import (
+from text_generation_server.layers import (
     TensorParallelColumnLinear,
     TensorParallelEmbedding,
     TensorParallelRowLinear,

server/text_generation_server/models/custom_modeling/opt_modeling.py

@@ -27,7 +27,7 @@ from transformers.modeling_outputs import (
 )
 from transformers.modeling_utils import PreTrainedModel
 from transformers import OPTConfig
-from text_generation_server.utils.layers import (
+from text_generation_server.layers import (
     FastLinear,
     TensorParallelColumnLinear,
     TensorParallelEmbedding,

server/text_generation_server/models/custom_modeling/phi_modeling.py

@@ -9,7 +9,7 @@ from typing import Optional, List, Tuple, Any
 from transformers.configuration_utils import PretrainedConfig
 from transformers.modeling_outputs import CausalLMOutputWithPast
 
-from text_generation_server.utils.layers import (
+from text_generation_server.layers import (
     TensorParallelRowLinear,
     TensorParallelColumnLinear,
     TensorParallelEmbedding,

server/text_generation_server/models/custom_modeling/t5_modeling.py

@@ -38,7 +38,7 @@ from transformers.utils import (
     is_torch_fx_proxy,
 )
 from transformers import T5Config
-from text_generation_server.utils.layers import (
+from text_generation_server.layers import (
     TensorParallelColumnLinear,
     TensorParallelEmbedding,
     TensorParallelRowLinear,

server/text_generation_server/models/flash_causal_lm.py

@@ -12,7 +12,6 @@ from dataclasses import dataclass
 from opentelemetry import trace
 from transformers import PreTrainedTokenizerBase
 from typing import Optional, Tuple, List, Type, Dict
-
 from text_generation_server.models import Model
 from text_generation_server.utils.tokens import batch_top_tokens
 from text_generation_server.utils.speculate import get_speculate
@@ -32,13 +31,14 @@ from text_generation_server.models.globals import MEM_POOL, CUDA_GRAPHS
 from text_generation_server.utils import StoppingCriteria, HeterogeneousNextTokenChooser
 from text_generation_server.utils.dist import MEMORY_FRACTION
 
-tracer = trace.get_tracer(__name__)
 from text_generation_server.utils.import_utils import (
-    IS_CUDA_SYSTEM,
+    empty_cache,
-    IS_ROCM_SYSTEM,
+    synchronize,
-    IS_XPU_SYSTEM,
+    get_free_memory,
 )
 
+tracer = trace.get_tracer(__name__)
+
 
 @dataclass
 class FlashCausalLMBatch(Batch):
@@ -757,10 +757,8 @@ class FlashCausalLM(Model):
 
     def warmup(self, batch: FlashCausalLMBatch):
         # The warmup batch is the biggest batch we could ever receive
-        if IS_CUDA_SYSTEM or IS_ROCM_SYSTEM:
+        empty_cache()
-            torch.cuda.empty_cache()
+
-        elif IS_XPU_SYSTEM:
-            torch.xpu.empty_cache()
         try:
             cache_manager = set_cache_manager(
                 batch.blocks,
@@ -780,10 +778,7 @@ class FlashCausalLM(Model):
                 f"You need to decrease `--max-batch-prefill-tokens`"
             ) from e
 
-        if IS_CUDA_SYSTEM or IS_ROCM_SYSTEM:
+        synchronize(self.device)
-            torch.cuda.synchronize(self.device)
-        elif IS_XPU_SYSTEM:
-            torch.xpu.synchronize(self.device)
 
         # Inspired by the original implementation in [vllm](https://github.com/vllm-project/vllm)
         # Calculate the number of blocks that can be allocated with the free memory
@@ -791,20 +786,7 @@ class FlashCausalLM(Model):
         cache_block_size = BLOCK_SIZE * self.num_kv_heads * self.head_size
         total_cache_size = self.num_layers * cache_block_size * 2 * dtype_size
 
-        if IS_CUDA_SYSTEM or IS_ROCM_SYSTEM:
+        free_memory = get_free_memory(self.device, MEMORY_FRACTION)
-            total_free_memory, _ = torch.cuda.mem_get_info(self.device)
-            total_gpu_memory = torch.cuda.get_device_properties(
-                self.device
-            ).total_memory
-
-            free_memory = max(
-                0, total_free_memory - (1 - MEMORY_FRACTION) * total_gpu_memory
-            )
-        elif IS_XPU_SYSTEM:
-            total_gpu_memory = torch.xpu.get_device_properties(self.device).total_memory
-            free_memory = int(total_gpu_memory * 0.5)
-        else:
-            raise NotImplementedError("FlashModel is only available on GPU")
 
         num_blocks = (
             # Leave 5% for some wiggle room

server/text_generation_server/models/flash_llama.py

@@ -18,7 +18,7 @@ from text_generation_server.utils import (
 
 tracer = trace.get_tracer(__name__)
 
-from text_generation_server.utils.import_utils import IS_XPU_SYSTEM
+from text_generation_server.utils.import_utils import SYSTEM
 
 
 class FlashLlama(FlashCausalLM):
@@ -35,7 +35,7 @@ class FlashLlama(FlashCausalLM):
         if torch.cuda.is_available():
             device = torch.device(f"cuda:{rank}")
             dtype = torch.float16 if dtype is None else dtype
-        elif IS_XPU_SYSTEM:
+        elif SYSTEM == "xpu":
             device = torch.device(f"xpu:{rank}")
             dtype = torch.float16 if dtype is None else dtype
         else:

server/text_generation_server/models/flash_mistral.py

@@ -33,7 +33,7 @@ tracer = trace.get_tracer(__name__)
 # Will be set in init
 SLIDING_WINDOW: Optional[int] = None
 SLIDING_WINDOW_BLOCKS: Optional[int] = None
-from text_generation_server.utils.import_utils import IS_XPU_SYSTEM
+from text_generation_server.utils.import_utils import SYSTEM
 
 MEM_POOL = torch.cuda.graph_pool_handle() if torch.cuda.is_available() else None
 
@@ -322,7 +322,7 @@ class BaseFlashMistral(FlashCausalLM):
         if torch.cuda.is_available():
             device = torch.device(f"cuda:{rank}")
             dtype = torch.float16 if dtype is None else dtype
-        elif IS_XPU_SYSTEM:
+        elif SYSTEM == "xpu":
             device = torch.device(f"xpu:{rank}")
             dtype = torch.float16 if dtype is None else dtype
         else:

server/text_generation_server/models/flash_neox.py

@@ -14,7 +14,7 @@ from text_generation_server.utils import (
     weight_files,
     Weights,
 )
-from text_generation_server.utils.import_utils import IS_XPU_SYSTEM
+from text_generation_server.utils.import_utils import SYSTEM
 
 tracer = trace.get_tracer(__name__)
 
@@ -33,7 +33,7 @@ class FlashNeoXSharded(FlashCausalLM):
         if torch.cuda.is_available():
             device = torch.device(f"cuda:{rank}")
             dtype = torch.float16 if dtype is None else dtype
-        elif IS_XPU_SYSTEM:
+        elif SYSTEM == "xpu":
             device = torch.device(f"xpu:{rank}")
             dtype = torch.float16 if dtype is None else dtype
         else:

server/text_generation_server/models/flash_rw.py

@@ -15,7 +15,7 @@ from text_generation_server.utils import (
     weight_files,
     Weights,
 )
-from text_generation_server.utils.import_utils import IS_XPU_SYSTEM
+from text_generation_server.utils.import_utils import SYSTEM
 
 tracer = trace.get_tracer(__name__)
 
@@ -34,7 +34,7 @@ class FlashRWSharded(FlashCausalLM):
         if torch.cuda.is_available():
             device = torch.device(f"cuda:{rank}")
             dtype = torch.float16 if dtype is None else dtype
-        elif IS_XPU_SYSTEM:
+        elif SYSTEM == "xpu":
             device = torch.device(f"xpu:{rank}")
             dtype = torch.float16 if dtype is None else dtype
         else:

server/text_generation_server/models/flash_santacoder.py

@@ -18,7 +18,7 @@ from text_generation_server.utils import (
     Weights,
 )
 
-from text_generation_server.utils.import_utils import IS_XPU_SYSTEM
+from text_generation_server.utils.import_utils import SYSTEM
 
 tracer = trace.get_tracer(__name__)
 
@@ -37,7 +37,7 @@ class FlashSantacoderSharded(FlashCausalLM):
         if torch.cuda.is_available():
             device = torch.device(f"cuda:{rank}")
             dtype = torch.float16 if dtype is None else dtype
-        elif IS_XPU_SYSTEM:
+        elif SYSTEM == "xpu":
             device = torch.device(f"xpu:{rank}")
             dtype = torch.float16 if dtype is None else dtype
         else:

server/text_generation_server/server.py

@@ -85,7 +85,7 @@ class TextGenerationService(generate_pb2_grpc.TextGenerationServiceServicer):
                 # When using GPTQ, Exllama kernels need some global kernels
                 # For which we have the finale shapes only after the model has loaded
                 # This will allocate those buffers.
-                from text_generation_server.utils.layers import (
+                from text_generation_server.layers.gptq import (
                     create_exllama_buffers,
                     set_device,
                 )

server/text_generation_server/utils/flash_attn.py

@@ -2,13 +2,8 @@ import os
 import torch
 
 from loguru import logger
-import math
 
-from text_generation_server.utils.import_utils import (
+from text_generation_server.utils.import_utils import SYSTEM
-    IS_CUDA_SYSTEM,
-    IS_ROCM_SYSTEM,
-    IS_XPU_SYSTEM,
-)
 
 if os.getenv("USE_FLASH_ATTENTION", "").lower() == "false":
     raise ImportError("`USE_FLASH_ATTENTION` is false.")
@@ -16,83 +11,22 @@ HAS_FLASH_ATTN = True
 HAS_FLASH_ATTN_V2_CUDA = False
 HAS_FLASH_ATTN_V2_ROCM = False
 
-if IS_XPU_SYSTEM:
+if SYSTEM == "xpu":
     import intel_extension_for_pytorch as ipex
 
-if IS_CUDA_SYSTEM or IS_ROCM_SYSTEM:
+    def attention(
-    if not torch.cuda.is_available():
+        q,
-        raise ImportError("CUDA is not available")
+        k,
+        v,
+        out,
+        cu_seqlens,
+        max_s,
+        softmax_scale,
+        window_size_left=-1,
+    ):
+        if window_size_left <= 0 and window_size_left != -1:
+            raise ValueError("`window_size_left` must be > 0 or -1")
 
-    major, minor = torch.cuda.get_device_capability()
-    is_sm75 = major == 7 and minor == 5
-    is_sm8x = major == 8 and minor >= 0
-    is_sm90 = major == 9 and minor == 0
-
-    HAS_FLASH_ATTN = False
-    HAS_FLASH_ATTN_V2_CUDA = False
-    HAS_FLASH_ATTN_V2_ROCM = False
-    try:
-        try:
-            import flash_attn_2_cuda
-        except ImportError:
-            architecture_suffix = ""
-            if IS_CUDA_SYSTEM:
-                architecture_suffix = "-cuda"
-            elif IS_ROCM_SYSTEM:
-                architecture_suffix = "-rocm"
-            raise ImportError(
-                "Flash Attention V2 is not installed.\n"
-                "Use the official Docker image (ghcr.io/huggingface/text-generation-inference:latest) "
-                f"or install flash attention v2 with `cd server && make install install-flash-attention-v2{architecture_suffix}`"
-            )
-        if not (is_sm8x or is_sm90):
-            raise ImportError(
-                f"GPU with CUDA capability {major} {minor} is not supported for "
-                "Flash Attention V2"
-            )
-        HAS_FLASH_ATTN_V2_CUDA = IS_CUDA_SYSTEM
-        HAS_FLASH_ATTN_V2_ROCM = IS_ROCM_SYSTEM
-    except ImportError as e:
-        try:
-            import flash_attn_cuda
-        except ImportError:
-            raise ImportError(
-                "Flash Attention is not installed.\n"
-                "Use the official Docker image (ghcr.io/huggingface/text-generation-inference:latest) "
-                "or install flash attention with `cd server && make install install-flash-attention`"
-            ) from e
-
-        if IS_CUDA_SYSTEM and not (is_sm75 or is_sm8x or is_sm90):
-            raise ImportError(
-                f"GPU with CUDA capability {major} {minor} is not supported"
-            ) from e
-        elif IS_ROCM_SYSTEM:
-            for idx in range(torch.cuda.device_count()):
-                if "MI210" not in torch.cuda.get_device_name(
-                    idx
-                ) and "MI250" not in torch.cuda.get_device_name(idx):
-                    raise ImportError(
-                        f"AMD GPU {torch.cuda.get_device_name(idx)} does not support flash-attention"
-                    )
-
-        logger.warning(f"Unable to use Flash Attention V2: {e}")
-        HAS_FLASH_ATTN = True
-
-
-def attention(
-    q,
-    k,
-    v,
-    out,
-    cu_seqlens,
-    max_s,
-    softmax_scale,
-    window_size_left=-1,
-):
-    if window_size_left <= 0 and window_size_left != -1:
-        raise ValueError("`window_size_left` must be > 0 or -1")
-
-    if IS_XPU_SYSTEM:
         if window_size_left != -1:
             raise ValueError(
                 f"XPU version of Flash Attention does not support window attention (window_size_left != -1, got window_size_left={window_size_left})."
@@ -114,7 +48,77 @@ def attention(
             None,
         )
 
-    if HAS_FLASH_ATTN_V2_CUDA:
+
+if SYSTEM in {"cuda", "rocm"}:
+    if not torch.cuda.is_available():
+        raise ImportError("CUDA is not available")
+
+    major, minor = torch.cuda.get_device_capability()
+    is_sm75 = major == 7 and minor == 5
+    is_sm8x = major == 8 and minor >= 0
+    is_sm90 = major == 9 and minor == 0
+
+    HAS_FLASH_ATTN = False
+    HAS_FLASH_ATTN_V2_CUDA = False
+    HAS_FLASH_ATTN_V2_ROCM = False
+    try:
+        try:
+            import flash_attn_2_cuda
+        except ImportError:
+            architecture_suffix = f"-{SYSTEM}"
+            raise ImportError(
+                "Flash Attention V2 is not installed.\n"
+                "Use the official Docker image (ghcr.io/huggingface/text-generation-inference:latest) "
+                f"or install flash attention v2 with `cd server && make install install-flash-attention-v2{architecture_suffix}`"
+            )
+        if not (is_sm8x or is_sm90):
+            raise ImportError(
+                f"GPU with CUDA capability {major} {minor} is not supported for "
+                "Flash Attention V2"
+            )
+        HAS_FLASH_ATTN_V2_CUDA = SYSTEM == "cuda"
+        HAS_FLASH_ATTN_V2_ROCM = SYSTEM == "rocm"
+    except ImportError as e:
+        try:
+            import flash_attn_cuda
+        except ImportError:
+            raise ImportError(
+                "Flash Attention is not installed.\n"
+                "Use the official Docker image (ghcr.io/huggingface/text-generation-inference:latest) "
+                "or install flash attention with `cd server && make install install-flash-attention`"
+            ) from e
+
+        if SYSTEM == "cuda" and not (is_sm75 or is_sm8x or is_sm90):
+            raise ImportError(
+                f"GPU with CUDA capability {major} {minor} is not supported"
+            ) from e
+        elif SYSTEM == "rocm":
+            for idx in range(torch.cuda.device_count()):
+                if "MI210" not in torch.cuda.get_device_name(
+                    idx
+                ) and "MI250" not in torch.cuda.get_device_name(idx):
+                    raise ImportError(
+                        f"AMD GPU {torch.cuda.get_device_name(idx)} does not support flash-attention"
+                    )
+
+        logger.warning(f"Unable to use Flash Attention V2: {e}")
+        HAS_FLASH_ATTN = True
+
+
+if HAS_FLASH_ATTN_V2_CUDA:
+
+    def attention(
+        q,
+        k,
+        v,
+        out,
+        cu_seqlens,
+        max_s,
+        softmax_scale,
+        window_size_left=-1,
+    ):
+        if window_size_left <= 0 and window_size_left != -1:
+            raise ValueError("`window_size_left` must be > 0 or -1")
         return flash_attn_2_cuda.varlen_fwd(
             q,
             k,
@@ -136,7 +140,21 @@ def attention(
             False,
             None,
         )
-    elif HAS_FLASH_ATTN_V2_ROCM:
+
+elif HAS_FLASH_ATTN_V2_ROCM:
+
+    def attention(
+        q,
+        k,
+        v,
+        out,
+        cu_seqlens,
+        max_s,
+        softmax_scale,
+        window_size_left=-1,
+    ):
+        if window_size_left <= 0 and window_size_left != -1:
+            raise ValueError("`window_size_left` must be > 0 or -1")
         if window_size_left != -1:
             raise ValueError(
                 f"RoCm version of Flash Attention v2 does not support window attention (window_size_left != -1, got window_size_left={window_size_left})."
@@ -159,7 +177,19 @@ def attention(
             False,
             None,
         )
-    elif HAS_FLASH_ATTN:
+
+elif HAS_FLASH_ATTN:
+
+    def attention(
+        q,
+        k,
+        v,
+        out,
+        cu_seqlens,
+        max_s,
+        softmax_scale,
+        window_size_left=-1,
+    ):
         if window_size_left != -1:
             raise NotImplementedError(
                 "window_size_left is only available with flash attn v2"
@@ -209,4 +239,5 @@ def attention(
             None,
         )
 
+else:
     raise NotImplementedError("flash attention is not installed")

server/text_generation_server/utils/gptq/quant_linear.py

@@ -1,359 +0,0 @@
-import math
-import numpy as np
-import torch
-import torch.nn as nn
-from torch.cuda.amp import custom_bwd, custom_fwd
-
-try:
-    import triton
-    import triton.language as tl
-    from . import custom_autotune
-
-    # code based https://github.com/fpgaminer/GPTQ-triton
-    @custom_autotune.autotune(
-        configs=[
-            triton.Config(
-                {
-                    "BLOCK_SIZE_M": 64,
-                    "BLOCK_SIZE_N": 256,
-                    "BLOCK_SIZE_K": 32,
-                    "GROUP_SIZE_M": 8,
-                },
-                num_stages=4,
-                num_warps=4,
-            ),
-            triton.Config(
-                {
-                    "BLOCK_SIZE_M": 128,
-                    "BLOCK_SIZE_N": 128,
-                    "BLOCK_SIZE_K": 32,
-                    "GROUP_SIZE_M": 8,
-                },
-                num_stages=4,
-                num_warps=4,
-            ),
-            triton.Config(
-                {
-                    "BLOCK_SIZE_M": 64,
-                    "BLOCK_SIZE_N": 128,
-                    "BLOCK_SIZE_K": 32,
-                    "GROUP_SIZE_M": 8,
-                },
-                num_stages=4,
-                num_warps=4,
-            ),
-            triton.Config(
-                {
-                    "BLOCK_SIZE_M": 128,
-                    "BLOCK_SIZE_N": 32,
-                    "BLOCK_SIZE_K": 32,
-                    "GROUP_SIZE_M": 8,
-                },
-                num_stages=4,
-                num_warps=4,
-            ),
-            triton.Config(
-                {
-                    "BLOCK_SIZE_M": 64,
-                    "BLOCK_SIZE_N": 64,
-                    "BLOCK_SIZE_K": 32,
-                    "GROUP_SIZE_M": 8,
-                },
-                num_stages=4,
-                num_warps=4,
-            ),
-            triton.Config(
-                {
-                    "BLOCK_SIZE_M": 64,
-                    "BLOCK_SIZE_N": 128,
-                    "BLOCK_SIZE_K": 32,
-                    "GROUP_SIZE_M": 8,
-                },
-                num_stages=2,
-                num_warps=8,
-            ),
-            triton.Config(
-                {
-                    "BLOCK_SIZE_M": 64,
-                    "BLOCK_SIZE_N": 64,
-                    "BLOCK_SIZE_K": 64,
-                    "GROUP_SIZE_M": 8,
-                },
-                num_stages=3,
-                num_warps=8,
-            ),
-            triton.Config(
-                {
-                    "BLOCK_SIZE_M": 32,
-                    "BLOCK_SIZE_N": 32,
-                    "BLOCK_SIZE_K": 128,
-                    "GROUP_SIZE_M": 8,
-                },
-                num_stages=2,
-                num_warps=4,
-            ),
-        ],
-        key=["M", "N", "K"],
-        nearest_power_of_two=True,
-        prune_configs_by={
-            "early_config_prune": custom_autotune.matmul248_kernel_config_pruner,
-            "perf_model": None,
-            "top_k": None,
-        },
-    )
-    @triton.jit
-    def matmul_248_kernel(
-        a_ptr,
-        b_ptr,
-        c_ptr,
-        scales_ptr,
-        zeros_ptr,
-        g_ptr,
-        M,
-        N,
-        K,
-        bits,
-        maxq,
-        stride_am,
-        stride_ak,
-        stride_bk,
-        stride_bn,
-        stride_cm,
-        stride_cn,
-        stride_scales,
-        stride_zeros,
-        BLOCK_SIZE_M: tl.constexpr,
-        BLOCK_SIZE_N: tl.constexpr,
-        BLOCK_SIZE_K: tl.constexpr,
-        GROUP_SIZE_M: tl.constexpr,
-    ):
-        """
-        Compute the matrix multiplication C = A x B.
-        A is of shape (M, K) float16
-        B is of shape (K//8, N) int32
-        C is of shape (M, N) float16
-        scales is of shape (G, N) float16
-        zeros is of shape (G, N) float16
-        g_ptr is of shape (K) int32
-        """
-        infearure_per_bits = 32 // bits
-
-        pid = tl.program_id(axis=0)
-        num_pid_m = tl.cdiv(M, BLOCK_SIZE_M)
-        num_pid_n = tl.cdiv(N, BLOCK_SIZE_N)
-        num_pid_k = tl.cdiv(K, BLOCK_SIZE_K)
-        num_pid_in_group = GROUP_SIZE_M * num_pid_n
-        group_id = pid // num_pid_in_group
-        first_pid_m = group_id * GROUP_SIZE_M
-        group_size_m = min(num_pid_m - first_pid_m, GROUP_SIZE_M)
-        pid_m = first_pid_m + (pid % group_size_m)
-        pid_n = (pid % num_pid_in_group) // group_size_m
-
-        offs_am = pid_m * BLOCK_SIZE_M + tl.arange(0, BLOCK_SIZE_M)
-        offs_bn = pid_n * BLOCK_SIZE_N + tl.arange(0, BLOCK_SIZE_N)
-        offs_k = tl.arange(0, BLOCK_SIZE_K)
-        a_ptrs = a_ptr + (
-            offs_am[:, None] * stride_am + offs_k[None, :] * stride_ak
-        )  # (BLOCK_SIZE_M, BLOCK_SIZE_K)
-        a_mask = offs_am[:, None] < M
-        # b_ptrs is set up such that it repeats elements along the K axis 8 times
-        b_ptrs = b_ptr + (
-            (offs_k[:, None] // infearure_per_bits) * stride_bk
-            + offs_bn[None, :] * stride_bn
-        )  # (BLOCK_SIZE_K, BLOCK_SIZE_N)
-        g_ptrs = g_ptr + offs_k
-        # shifter is used to extract the N bits of each element in the 32-bit word from B
-        scales_ptrs = scales_ptr + offs_bn[None, :]
-        zeros_ptrs = zeros_ptr + (offs_bn[None, :] // infearure_per_bits)
-
-        shifter = (offs_k % infearure_per_bits) * bits
-        zeros_shifter = (offs_bn % infearure_per_bits) * bits
-        accumulator = tl.zeros((BLOCK_SIZE_M, BLOCK_SIZE_N), dtype=tl.float32)
-
-        for k in range(0, num_pid_k):
-            g_idx = tl.load(g_ptrs)
-
-            # Fetch scales and zeros; these are per-outfeature and thus reused in the inner loop
-            scales = tl.load(
-                scales_ptrs + g_idx[:, None] * stride_scales
-            )  # (BLOCK_SIZE_K, BLOCK_SIZE_N,)
-            zeros = tl.load(
-                zeros_ptrs + g_idx[:, None] * stride_zeros
-            )  # (BLOCK_SIZE_K, BLOCK_SIZE_N,)
-
-            zeros = (zeros >> zeros_shifter[None, :]) & maxq
-            zeros = (zeros + 1) & maxq  # eventually avoid overflow
-
-            a = tl.load(a_ptrs, mask=a_mask, other=0.0)  # (BLOCK_SIZE_M, BLOCK_SIZE_K)
-            b = tl.load(b_ptrs)  # (BLOCK_SIZE_K, BLOCK_SIZE_N), but repeated
-
-            # Now we need to unpack b (which is N-bit values) into 32-bit values
-            b = (b >> shifter[:, None]) & maxq  # Extract the N-bit values
-            b = (b - zeros) * scales  # Scale and shift
-
-            accumulator += tl.dot(a, b)
-            a_ptrs += BLOCK_SIZE_K
-            b_ptrs += (BLOCK_SIZE_K // infearure_per_bits) * stride_bk
-            g_ptrs += BLOCK_SIZE_K
-
-        c_ptrs = c_ptr + stride_cm * offs_am[:, None] + stride_cn * offs_bn[None, :]
-        c_mask = (offs_am[:, None] < M) & (offs_bn[None, :] < N)
-        tl.store(c_ptrs, accumulator, mask=c_mask)
-
-except:
-    print("triton not installed.")
-
-
-def matmul248(input, qweight, scales, qzeros, g_idx, bits, maxq):
-    with torch.cuda.device(input.device):
-        output = torch.empty(
-            (input.shape[0], qweight.shape[1]), device=input.device, dtype=torch.float16
-        )
-        grid = lambda META: (
-            triton.cdiv(input.shape[0], META["BLOCK_SIZE_M"])
-            * triton.cdiv(qweight.shape[1], META["BLOCK_SIZE_N"]),
-        )
-        matmul_248_kernel[grid](
-            input,
-            qweight,
-            output,
-            scales,
-            qzeros,
-            g_idx,
-            input.shape[0],
-            qweight.shape[1],
-            input.shape[1],
-            bits,
-            maxq,
-            input.stride(0),
-            input.stride(1),
-            qweight.stride(0),
-            qweight.stride(1),
-            output.stride(0),
-            output.stride(1),
-            scales.stride(0),
-            qzeros.stride(0),
-        )
-        return output
-
-
-class QuantLinearFunction(torch.autograd.Function):
-    @staticmethod
-    @custom_fwd(cast_inputs=torch.float16)
-    def forward(ctx, input, qweight, scales, qzeros, g_idx, bits, maxq):
-        output = matmul248(input, qweight, scales, qzeros, g_idx, bits, maxq)
-        return output
-
-
-class QuantLinear(nn.Module):
-    def __init__(self, qweight, qzeros, scales, g_idx, bias, bits, groupsize):
-        super().__init__()
-        self.register_buffer("qweight", qweight)
-        self.register_buffer("qzeros", qzeros)
-        self.register_buffer("scales", scales)
-        self.register_buffer("g_idx", g_idx)
-        if bias is not None:
-            self.register_buffer("bias", bias)
-        else:
-            self.bias = None
-        if bits not in [2, 4, 8]:
-            raise NotImplementedError("Only 2,4,8 bits are supported.")
-        self.bits = bits
-        self.maxq = 2**self.bits - 1
-        self.groupsize = groupsize
-
-        self.outfeatures = qweight.shape[1]
-        self.infeatures = qweight.shape[0] * 32 // bits
-
-    @classmethod
-    def new(cls, bits, groupsize, infeatures, outfeatures, bias):
-        if bits not in [2, 4, 8]:
-            raise NotImplementedError("Only 2,4,8 bits are supported.")
-
-        qweight = torch.zeros((infeatures // 32 * bits, outfeatures), dtype=torch.int32)
-        qzeros = torch.zeros(
-            (math.ceil(infeatures / groupsize), outfeatures // 32 * bits),
-            dtype=torch.int32,
-        )
-        scales = torch.zeros(
-            (math.ceil(infeatures / groupsize), outfeatures), dtype=torch.float16
-        )
-        g_idx = torch.tensor(
-            [i // groupsize for i in range(infeatures)], dtype=torch.int32
-        )
-        if bias:
-            bias = torch.zeros((outfeatures), dtype=torch.float16)
-        else:
-            bias = None
-        return cls(qweight, qzeros, scales, g_idx, bias, bits, groupsize)
-
-    def pack(self, linear, scales, zeros, g_idx=None):
-        self.g_idx = g_idx.clone() if g_idx is not None else self.g_idx
-
-        scales = scales.t().contiguous()
-        zeros = zeros.t().contiguous()
-        scale_zeros = zeros * scales
-        self.scales = scales.clone().half()
-        if linear.bias is not None:
-            self.bias = linear.bias.clone().half()
-
-        intweight = []
-        for idx in range(self.infeatures):
-            intweight.append(
-                torch.round(
-                    (linear.weight.data[:, idx] + scale_zeros[self.g_idx[idx]])
-                    / self.scales[self.g_idx[idx]]
-                ).to(torch.int)[:, None]
-            )
-        intweight = torch.cat(intweight, dim=1)
-        intweight = intweight.t().contiguous()
-        intweight = intweight.numpy().astype(np.uint32)
-        qweight = np.zeros(
-            (intweight.shape[0] // 32 * self.bits, intweight.shape[1]), dtype=np.uint32
-        )
-        i = 0
-        row = 0
-        while row < qweight.shape[0]:
-            if self.bits in [2, 4, 8]:
-                for j in range(i, i + (32 // self.bits)):
-                    qweight[row] |= intweight[j] << (self.bits * (j - i))
-                i += 32 // self.bits
-                row += 1
-            else:
-                raise NotImplementedError("Only 2,4,8 bits are supported.")
-
-        qweight = qweight.astype(np.int32)
-        self.qweight = torch.from_numpy(qweight)
-
-        zeros -= 1
-        zeros = zeros.numpy().astype(np.uint32)
-        qzeros = np.zeros(
-            (zeros.shape[0], zeros.shape[1] // 32 * self.bits), dtype=np.uint32
-        )
-        i = 0
-        col = 0
-        while col < qzeros.shape[1]:
-            if self.bits in [2, 4, 8]:
-                for j in range(i, i + (32 // self.bits)):
-                    qzeros[:, col] |= zeros[:, j] << (self.bits * (j - i))
-                i += 32 // self.bits
-                col += 1
-            else:
-                raise NotImplementedError("Only 2,4,8 bits are supported.")
-
-        qzeros = qzeros.astype(np.int32)
-        self.qzeros = torch.from_numpy(qzeros)
-
-    def forward(self, x):
-        out_shape = x.shape[:-1] + (self.outfeatures,)
-        out = QuantLinearFunction.apply(
-            x.reshape(-1, x.shape[-1]),
-            self.qweight,
-            self.scales,
-            self.qzeros,
-            self.g_idx,
-            self.bits,
-            self.maxq,
-        )
-        out = out + self.bias if self.bias is not None else out
-        return out.reshape(out_shape)

server/text_generation_server/utils/import_utils.py

@@ -10,6 +10,41 @@ def is_xpu_available():
     return hasattr(torch, "xpu") and torch.xpu.is_available()
 
 
-IS_ROCM_SYSTEM = torch.version.hip is not None
+def get_cuda_free_memory(device, memory_fraction):
-IS_CUDA_SYSTEM = torch.version.cuda is not None
+    total_free_memory, _ = torch.cuda.mem_get_info(device)
-IS_XPU_SYSTEM = is_xpu_available()
+    total_gpu_memory = torch.cuda.get_device_properties(device).total_memory
+    free_memory = max(0, total_free_memory - (1 - memory_fraction) * total_gpu_memory)
+    return free_memory
+
+
+def get_xpu_free_memory(device):
+    total_gpu_memory = torch.xpu.get_device_properties(device).total_memory
+    free_memory = int(total_gpu_memory * 0.5)
+    return free_memory
+
+
+SYSTEM = None
+if torch.version.hip is not None:
+    SYSTEM = "rocm"
+    empty_cache = torch.cuda.empty_cache
+    synchronize = torch.cuda.synchronize
+    get_free_memory = get_cuda_free_memory
+elif torch.version.cuda is not None and torch.cuda.is_available():
+    SYSTEM = "cuda"
+    empty_cache = torch.cuda.empty_cache
+    synchronize = torch.cuda.synchronize
+    get_free_memory = get_cuda_free_memory
+elif is_xpu_available():
+    SYSTEM = "xpu"
+    empty_cache = torch.xpu.empty_cache
+    synchronize = torch.xpu.synchronize
+    get_free_memory = get_xpu_free_memory
+else:
+    SYSTEM = "cpu"
+
+    def noop(*args, **kwargs):
+        pass
+
+    empty_cache = noop
+    synchronize = noop
+    get_free_memory = noop

server/text_generation_server/utils/paged_attention.py

@@ -1,13 +1,9 @@
 import torch
-from text_generation_server.utils.import_utils import (
+from text_generation_server.utils.import_utils import SYSTEM
-    IS_CUDA_SYSTEM,
-    IS_ROCM_SYSTEM,
-    IS_XPU_SYSTEM,
-)
 
 _PARTITION_SIZE = 512
 
-if IS_XPU_SYSTEM:
+if SYSTEM == "xpu":
     import intel_extension_for_pytorch as ipex
 
 
@@ -18,17 +14,17 @@ def reshape_and_cache(
     value_cache: torch.Tensor,
     slots: torch.Tensor,
 ):
-    if IS_CUDA_SYSTEM:
+    if SYSTEM == "cuda":
         from vllm._C import cache_ops
 
         cache_ops.reshape_and_cache(
             key, value, key_cache, value_cache, slots, "auto", 1.0
         )
-    elif IS_ROCM_SYSTEM:
+    elif SYSTEM == "rocm":
         from vllm import cache_ops
 
         cache_ops.reshape_and_cache(key, value, key_cache, value_cache, slots)
-    elif IS_XPU_SYSTEM:
+    elif SYSTEM == "xpu":
         ipex.llm.modules.PagedAttention.reshape_and_cache(
             key, value, key_cache, value_cache, slots
         )
@@ -68,7 +64,7 @@ def attention(
     block_size = value_cache.shape[3]
     num_seqs, num_heads, head_size = query.shape
     max_num_partitions = (max_s + _PARTITION_SIZE - 1) // _PARTITION_SIZE
-    if IS_XPU_SYSTEM:
+    if SYSTEM == "xpu":
         query = query.contiguous()
         return ipex.llm.modules.PagedAttention.single_query_cached_kv_attention(
             out,
@@ -91,7 +87,7 @@ def attention(
     # to parallelize.
     use_v1 = max_s <= 8192 and (max_num_partitions == 1 or num_seqs * num_heads > 512)
     if use_v1:
-        if IS_CUDA_SYSTEM:
+        if SYSTEM == "cuda":
             from vllm._C import ops
 
             ops.paged_attention_v1(
@@ -109,7 +105,7 @@ def attention(
                 "auto",
                 1.0,
             )
-        elif IS_ROCM_SYSTEM:
+        elif SYSTEM == "rocm":
             from vllm import attention_ops
 
             attention_ops.paged_attention_v1(
@@ -143,7 +139,7 @@ def attention(
         )
         max_logits = torch.empty_like(exp_sums)
 
-        if IS_CUDA_SYSTEM:
+        if SYSTEM == "cuda":
             from vllm._C import ops
 
             ops.paged_attention_v2(
@@ -164,7 +160,7 @@ def attention(
                 "auto",
                 1.0,
             )
-        elif IS_ROCM_SYSTEM:
+        elif SYSTEM == "rocm":
             from vllm import attention_ops
 
             attention_ops.paged_attention_v2(

server/text_generation_server/utils/weights.py

@@ -171,7 +171,7 @@ class Weights:
                 log_once(
                     logger.info, "Converting AWQ model to Exllama/GPTQ packing format."
                 )
-                from text_generation_server.utils.awq.conversion_utils import (
+                from text_generation_server.layers.awq.conversion_utils import (
                     fast_awq_to_gptq,
                 )
 
@@ -227,7 +227,7 @@ class Weights:
 
             bits, groupsize, desc_act, quant_method = self._get_gptq_params()
 
-            from text_generation_server.utils.layers import HAS_EXLLAMA
+            from text_generation_server.layers.gptq import HAS_EXLLAMA
 
             use_exllama = (
                 bits == 4 and HAS_EXLLAMA and quantize == "gptq" and not desc_act
@@ -242,7 +242,7 @@ class Weights:
                 log_once(
                     logger.info, "Converting AWQ model to Exllama/GPTQ packing format."
                 )
-                from text_generation_server.utils.awq.conversion_utils import (
+                from text_generation_server.layers.awq.conversion_utils import (
                     fast_awq_to_gptq,
                 )
 
@@ -321,7 +321,7 @@ class Weights:
                         # it would require to reorder input activations that are split unto several GPUs
                         use_exllama = False
 
-            from text_generation_server.utils.layers import HAS_EXLLAMA, CAN_EXLLAMA
+            from text_generation_server.layers.gptq import HAS_EXLLAMA, CAN_EXLLAMA
 
             if use_exllama:
                 if not HAS_EXLLAMA:
@@ -348,7 +348,7 @@ class Weights:
                 log_once(
                     logger.info, "Converting AWQ model to Exllama/GPTQ packing format."
                 )
-                from text_generation_server.utils.awq.conversion_utils import (
+                from text_generation_server.layers.awq.conversion_utils import (
                     fast_awq_to_gptq,
                 )