[MXFP4] Add calibration support #509
Open
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This was referenced Nov 4, 2025
Signed-off-by: Dipika Sikka <dipikasikka1@gmail.com>
This was referenced Nov 17, 2025
dsikka
added a commit
to vllm-project/llm-compressor
that referenced
this pull request
Nov 17, 2025
# Summary - Requires: vllm-project/compressed-tensors#509 - Add script to generate an mxfp4 quantized model - This feature is currently experimental as support has not landed or tested in vLLM # Testing: Sample Model: - nm-testing/Meta-Llama-3-8B-Instruct-MXFP4 Sample Generation (Transformers): ```bash ========== SAMPLE GENERATION ============== <|begin_of_text|>Hello my name is Sophia and I am a 3rd year student at the University of California, Berkeley. I am a double major in Linguistics and Psychology, with a minor in Education. I am very interested in the way that language and culture interact, and I believe that education is the key to creating a more just and equitable society. I am a native speaker of English, and I have also studied Spanish, French, and Mandarin Chinese. I am very interested in the way that language can be used to bring ========================================== ``` Sample Config: ```yaml "quantization_config": { "config_groups": { "group_0": { "format": "mxfp4-pack-quantized", "input_activations": { "actorder": null, "block_structure": null, "dynamic": true, "group_size": 32, "num_bits": 4, "observer": null, "observer_kwargs": {}, "scale_dtype": "torch.uint8", "strategy": "group", "symmetric": true, "type": "float", "zp_dtype": null }, "output_activations": null, "targets": [ "Linear" ], "weights": { "actorder": null, "block_structure": null, "dynamic": false, "group_size": 32, "num_bits": 4, "observer": "minmax", "observer_kwargs": {}, "scale_dtype": "torch.uint8", "strategy": "group", "symmetric": true, "type": "float", "zp_dtype": null } } }, "format": "mxfp4-pack-quantized", } ``` --------- Signed-off-by: Dipika Sikka <ds3822@columbia.edu> Co-authored-by: gemini-code-assist[bot] <176961590+gemini-code-assist[bot]@users.noreply.github.com>
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Summary
mxfp4_packed_compressorto additionally compress the generated scales during compression time. Decompression is not yet supported as general decompression of qparams does not work right now, until this fix lands: [WIP] fix qparams decompression #514Testing: