[Kernel][FP8] Initial support with dynamic per-tensor scaling (#4118)

Provide an initial support to FP8 computation. This PR is inspired by HuggingFace TGI: huggingface/text-generation-inference#1726

This feature can be enabled with --quantization fp8 or -q fp8 when launching an engine.

Algorithm:
We still load a model checkpoint in FP16/BF16. After the weights are loaded, Fp8LinearMethod calculates the per-tensor scaling factor of weights and quantizes the weights accordingly. The scaling factor will then be stored for future use. Meanwhile, the per-tensor scaling factor for activations is calculated in every forward pass.

Initial Results:
Currently tested Mistral-7B on 1xH100. With prompt length ~5 and decoding length 128:

BF16: 1.47s
FP8: 1.66s
I'll try to use larger models and try to find more performance bottleneck. Meanwhile, you're welcome to try this code.
This commit is contained in:
Cody Yu
2024-04-20 04:28:57 +00:00
committed by GitHub
parent 682789d402
commit a22cdea371
7 changed files with 189 additions and 5 deletions
+5 -4
View File
@@ -42,10 +42,11 @@ class LLM:
However, if the `torch_dtype` in the config is `float32`, we will
use `float16` instead.
quantization: The method used to quantize the model weights. Currently,
we support "awq", "gptq" and "squeezellm". If None, we first check
the `quantization_config` attribute in the model config file. If
that is None, we assume the model weights are not quantized and use
`dtype` to determine the data type of the weights.
we support "awq", "gptq", "squeezellm", and "fp8" (experimental).
If None, we first check the `quantization_config` attribute in the
model config file. If that is None, we assume the model weights are
not quantized and use `dtype` to determine the data type of
the weights.
revision: The specific model version to use. It can be a branch name,
a tag name, or a commit id.
tokenizer_revision: The specific tokenizer version to use. It can be a