Skip to content

GitLab

"tests/vscode:/vscode.git/clone" did not exist on "bfdb1ba5c3fb14387c69acb1f5067102d8028e56"
Switch branch/tag
  1. 25 May, 2024 1 commit
  3. 23 May, 2024 1 commit
  5. 22 May, 2024 1 commit
  7. 16 May, 2024 2 commits
  9. 03 May, 2024 1 commit
  11. 29 Apr, 2024 1 commit
  13. 27 Apr, 2024 1 commit
  15. 24 Apr, 2024 1 commit
    • Philipp Moritz's avatar
      [Kernel] FP8 support for MoE kernel / Mixtral (#4244) · eace8bf0
      Philipp Moritz authored 
      This PR is the first step towards fixing https://github.com/vllm-project/vllm/pull/3208

It implements dynamic per-tensor scaling (see https://github.com/vllm-project/vllm/pull/4118), so users do not need to compute activation scales on a calibration dataset and they also don't need to convert their model checkpoints. It is enough to specify the `quantization="fp8"` argument. You can try out the PR like this:

```python
from vllm import LLM, SamplingParams

prompts = [
    "Hello, my name is",
    "The president of the United States is",
    "The capital of France is",
    "The future of AI is",
]
sampling_params = SamplingParams(temperature=0.8, top_p=0.95)

llm = LLM(model="mistralai/Mixtral-8x7B-Instruct-v0.1", tensor_parallel_size=2, quantization="fp8")

outputs = llm.generate(prompts, sampling_params)

# Print the outputs.
for output in outputs:
    prompt = output.prompt
    generated_text = output.outputs[0].text
    print(f"Prompt: {prompt!r}, Generated text: {generated_text!r}")
```

**Performance**: For this PR, the focus is on making the code clean (while still trying to get reasonable performance), there is a bunch of optimizations that we will submit as a follow up PR that significantly improve the performance (similar to the numbers in https://github.com/vllm-project/vllm/pull/3954). With this PR, the results are as follows:

<img width="725" alt="Screenshot 2024-04-21 at 1 31 50 PM" src="https://github.com/vllm-project/vllm/assets/113316/d8fe1118-07a0-4d4e-8530-37a77d465a03">


**Accuracy**: The accuracy with this PR on MMLU on `mistralai/Mixtral-8x7B-v0.1` is as follows:

```
|      Groups      |Version|Filter|n-shot|Metric|Value |   |Stderr|
|------------------|-------|------|-----:|------|-----:|---|-----:|
|mmlu              |N/A    |none  |     0|acc   |0.7018|±  |0.0036|
| - humanities     |N/A    |none  |     5|acc   |0.6472|±  |0.0065|
| - other          |N/A    |none  |     5|acc   |0.7673|±  |0.0072|
| - social_sciences|N/A    |none  |     5|acc   |0.8099|±  |0.0070|
| - stem           |N/A    |none  |     5|acc   |0.6131|±  |0.0083|
```
this compares favorably with the fp16 results which are
```
|      Groups      |Version|Filter|n-shot|Metric|Value |   |Stderr|
|------------------|-------|------|-----:|------|-----:|---|-----:|
|mmlu              |N/A    |none  |     0|acc   |0.7020|±  |0.1313|
| - humanities     |N/A    |none  |     5|acc   |0.6425|±  |0.1349|
| - other          |N/A    |none  |     5|acc   |0.7744|±  |0.1038|
| - social_sciences|N/A    |none  |     5|acc   |0.8131|±  |0.0695|
| - stem           |N/A    |none  |     5|acc   |0.6108|±  |0.1383|
```

Happy hacking!
      eace8bf0
  17. 23 Apr, 2024 1 commit
  19. 03 Apr, 2024 1 commit
  21. 13 Mar, 2024 2 commits
  23. 01 Mar, 2024 1 commit
  25. 26 Feb, 2024 1 commit
  27. 22 Feb, 2024 1 commit
  29. 06 Feb, 2024 1 commit
  31. 30 Jan, 2024 2 commits
  33. 29 Jan, 2024 1 commit
  35. 27 Jan, 2024 2 commits
  37. 26 Jan, 2024 1 commit
  39. 15 Dec, 2023 1 commit
  41. 08 Dec, 2023 1 commit
  43. 24 Nov, 2023 1 commit