enable bfloat16 for cpu (#1034)

if there's no cuda. disable custom kernels

# 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

 -->

Signed-off-by: Wang, Yi A <yi.a.wang@intel.com>

server/text_generation_server/models/__init__.py

@@ -153,7 +153,7 @@ def get_model(
         )
     elif model_type == "mpt":
         return MPTSharded(
-            model_id, revision, quantize=quantize, trust_remote_code=trust_remote_code
+            model_id, revision, quantize=quantize, dtype=dtype, trust_remote_code=trust_remote_code
         )
 
     elif model_type == "gpt_neox":

server/text_generation_server/models/bloom.py

@@ -51,7 +51,7 @@ class BLOOMSharded(CausalLM):
             dtype = torch.float16 if dtype is None else dtype
         else:
             device = torch.device("cpu")
-            dtype = torch.float32
+            dtype = torch.float32 if dtype is None else dtype
 
         tokenizer = AutoTokenizer.from_pretrained(
             model_id,

server/text_generation_server/models/causal_lm.py

@@ -492,7 +492,7 @@ class CausalLM(Model):
                 raise ValueError("quantization is not available on CPU")
 
             device = torch.device("cpu")
-            dtype = torch.float32
+            dtype = torch.float32 if dtype is None else dtype
 
         tokenizer = AutoTokenizer.from_pretrained(
             model_id,

server/text_generation_server/models/custom_modeling/bloom_modeling.py

@@ -40,7 +40,7 @@ from text_generation_server.utils.layers import (
 )
 
 CUSTOM_KERNELS_ENABLED = False
-if not os.environ.get("DISABLE_CUSTOM_KERNELS", "False") == "True":
+if torch.cuda.is_available() and not os.environ.get("DISABLE_CUSTOM_KERNELS", "False") == "True":
     try:
         from custom_kernels import fused_bloom_attention_cuda
 

server/text_generation_server/models/custom_modeling/neox_modeling.py

@@ -49,7 +49,7 @@ from text_generation_server.utils.layers import (
 
 
 CUSTOM_KERNELS_ENABLED = False
-if not os.environ.get("DISABLE_CUSTOM_KERNELS", "False") == "True":
+if torch.cuda.is_available() and not os.environ.get("DISABLE_CUSTOM_KERNELS", "False") == "True":
     try:
         from custom_kernels import fused_attention_cuda
 

server/text_generation_server/models/galactica.py

@@ -167,7 +167,7 @@ class GalacticaSharded(CausalLM):
             dtype = torch.float16 if dtype is None else dtype
         else:
             device = torch.device("cpu")
-            dtype = torch.float32
+            dtype = torch.float32 if dtype is None else dtype
 
         tokenizer = AutoTokenizer.from_pretrained(
             model_id,

server/text_generation_server/models/gpt_neox.py

@@ -33,7 +33,7 @@ class GPTNeoxSharded(CausalLM):
             dtype = torch.float16 if dtype is None else dtype
         else:
             device = torch.device("cpu")
-            dtype = torch.float32
+            dtype = torch.float32 if dtype is None else dtype
 
         tokenizer = AutoTokenizer.from_pretrained(
             model_id,

server/text_generation_server/models/idefics.py

@@ -42,7 +42,7 @@ class IDEFICSSharded(IdeficsCausalLM):
             dtype = torch.bfloat16 if dtype is None else dtype
         else:
             device = torch.device("cpu")
-            dtype = torch.float32
+            dtype = torch.float32 if dtype is None else dtype
         self.device, self.dtype = device, dtype
 
         config = IdeficsConfig.from_pretrained(

server/text_generation_server/models/idefics_causal_lm.py

@@ -560,7 +560,7 @@ class IdeficsCausalLM(Model):
                 raise ValueError("quantization is not available on CPU")
 
             device = torch.device("cpu")
-            dtype = torch.float32
+            dtype = torch.float32 if dtype is None else dtype
 
         tokenizer = AutoTokenizer.from_pretrained(
             model_id,

server/text_generation_server/models/mpt.py

@@ -43,14 +43,16 @@ class MPTSharded(CausalLM):
         model_id: str,
         revision: Optional[str] = None,
         quantize: Optional[str] = None,
+        dtype: Optional[torch.dtype] = None,
         trust_remote_code: bool = False,
     ):
         self.process_group, rank, world_size = initialize_torch_distributed()
         if torch.cuda.is_available():
             device = torch.device(f"cuda:{rank}")
-            dtype = torch.float16
+            dtype = torch.float16 if dtype is None else dtype
         else:
-            raise NotImplementedError("MPTSharded is only available on GPU")
+            device = torch.device("cpu")
+            dtype = torch.float32 if dtype is None else dtype
 
         tokenizer = AutoTokenizer.from_pretrained(
             model_id,

server/text_generation_server/models/opt.py

@@ -31,7 +31,7 @@ class OPTSharded(CausalLM):
             dtype = torch.float16 if dtype is None else dtype
         else:
             device = torch.device("cpu")
-            dtype = torch.float32
+            dtype = torch.float32 if dtype is None else dtype
 
         tokenizer = AutoTokenizer.from_pretrained(
             model_id,

server/text_generation_server/models/rw.py

@@ -23,7 +23,7 @@ class RW(CausalLM):
                 raise ValueError("quantization is not available on CPU")
 
             device = torch.device("cpu")
-            dtype = torch.float32
+            dtype = torch.float32 if dtype is None else dtype
 
         tokenizer = AutoTokenizer.from_pretrained(
             model_id,

server/text_generation_server/models/santacoder.py

@@ -30,7 +30,7 @@ class SantaCoder(CausalLM):
                 raise ValueError("quantization is not available on CPU")
 
             device = torch.device("cpu")
-            dtype = torch.float32
+            dtype = torch.float32 if dtype is None else dtype
 
         tokenizer = AutoTokenizer.from_pretrained(
             model_id,

server/text_generation_server/models/seq2seq_lm.py

@@ -541,7 +541,7 @@ class Seq2SeqLM(Model):
                 raise ValueError("quantization is not available on CPU")
 
             device = torch.device("cpu")
-            dtype = torch.float32
+            dtype = torch.float32 if dtype is None else dtype
 
         model = AutoModelForSeq2SeqLM.from_pretrained(
             model_id,

server/text_generation_server/models/t5.py

@@ -34,7 +34,7 @@ class T5Sharded(Seq2SeqLM):
             dtype = torch.float16 if dtype is None else dtype
         else:
             device = torch.device("cpu")
-            dtype = torch.float32
+            dtype = torch.float32 if dtype is None else dtype
 
         config = AutoConfig.from_pretrained(
             model_id,