Mirrors
/
hf_text-generation-inference
mirror of https://github.com/huggingface/text-generation-inference.git
1
0
Fork 0
Code Issues Packages Projects Releases Wiki Activity
hf_text-generation-inference/server
History
Daniël de Kok 3c9df21ff8
Add support for compressed-tensors w8a8 int checkpoints (#2745) 
* Add support for compressed-tensors w8a8 int checkpoints

This change adds a loader for w8a8 int checkpoints. One large benefit of
int8 support is that the corresponding cutlass matmul kernels also work on
compute capability 7.5.

Evaluation on neuralmagic/Meta-Llama-3.1-8B-Instruct-quantized.w8a8:

|     Tasks     |Version|     Filter     |n-shot|        Metric         |   |Value |   |Stderr|
|---------------|------:|----------------|-----:|-----------------------|---|-----:|---|------|
|gsm8k_cot_llama|      3|flexible-extract|     8|exact_match            |↑  |0.8431|±  |0.0100|
|               |       |strict-match    |     8|exact_match            |↑  |0.8393|±  |0.0101|
|ifeval         |      4|none            |     0|inst_level_loose_acc   |↑  |0.8597|±  |   N/A|
|               |       |none            |     0|inst_level_strict_acc  |↑  |0.8201|±  |   N/A|
|               |       |none            |     0|prompt_level_loose_acc |↑  |0.7967|±  |0.0173|
|               |       |none            |     0|prompt_level_strict_acc|↑  |0.7468|±  |0.0187|

Which is the same ballpark as vLLM.

As usual, lots of thanks to Neural Magic/vLLM for the kernels.

* Always use dynamic input quantization for w8a8 int

It's far less flaky and gives better output.

* Use marlin-kernels 0.3.5

* Fix a typo

Co-authored-by: drbh <david.richard.holtz@gmail.com>

* Small fixes

---------

Co-authored-by: drbh <david.richard.holtz@gmail.com>
 		2024-11-18 17:20:31 +01:00
..
custom_kernels All integration tests back everywhere (too many failed CI). (#2428) 2024-08-16 21:19:46 +02:00
exllama_kernels Update ROCM libs and improvements (#2579) 2024-09-30 10:54:32 +02:00
exllamav2_kernels Update ROCM libs and improvements (#2579) 2024-09-30 10:54:32 +02:00
tests feat: prefill chunking (#2600) 2024-10-16 12:49:33 +02:00
text_generation_server Add support for compressed-tensors w8a8 int checkpoints (#2745) 2024-11-18 17:20:31 +01:00
.gitignore Impl simple mamba model (#1480) 2024-02-08 10:19:45 +01:00
Makefile Remove vLLM dependency for CUDA (#2751) 2024-11-17 17:34:50 +01:00
Makefile-awq chore: add pre-commit (#1569) 2024-02-16 11:58:58 +01:00
Makefile-eetq Upgrade EETQ (Fixes the cuda graphs). (#1729) 2024-04-12 08:15:28 +02:00
Makefile-exllamav2 Upgrading exl2. (#2415) 2024-08-14 11:58:08 +02:00
Makefile-flash-att Hotfixing `make install`. (#2008) 2024-06-04 23:34:03 +02:00
Makefile-flash-att-v2 Update ROCM libs and improvements (#2579) 2024-09-30 10:54:32 +02:00
Makefile-flashinfer Prefix test - Different kind of load test to trigger prefix test bugs. (#2490) 2024-09-11 18:10:40 +02:00
Makefile-lorax-punica Enable multiple LoRa adapters (#2010) 2024-06-25 14:46:27 -04:00
Makefile-selective-scan chore: add pre-commit (#1569) 2024-02-16 11:58:58 +01:00
Makefile-vllm Remove vLLM dependency for CUDA (#2751) 2024-11-17 17:34:50 +01:00
README.md chore: add pre-commit (#1569) 2024-02-16 11:58:58 +01:00
poetry.lock Add support for compressed-tensors w8a8 int checkpoints (#2745) 2024-11-18 17:20:31 +01:00
pyproject.toml Add support for compressed-tensors w8a8 int checkpoints (#2745) 2024-11-18 17:20:31 +01:00
requirements_cuda.txt Upgrading our deps. (#2750) 2024-11-15 14:03:27 +01:00
requirements_intel.txt Upgrading our deps. (#2750) 2024-11-15 14:03:27 +01:00
requirements_rocm.txt Upgrading our deps. (#2750) 2024-11-15 14:03:27 +01:00

README.md

Text Generation Inference Python gRPC Server

A Python gRPC server for Text Generation Inference

Install

make install

Run

make run-dev