[CT] Fix CT Config to honor fp8_inc KV cache dtype#929
Merged
xuechendi merged 3 commits intovllm-project:mainfrom Feb 5, 2026
Merged
[CT] Fix CT Config to honor fp8_inc KV cache dtype#929xuechendi merged 3 commits intovllm-project:mainfrom
fp8_inc KV cache dtype#929xuechendi merged 3 commits intovllm-project:mainfrom
Conversation
Signed-off-by: yiliu30 <yi4.liu@intel.com>
yiliu30
requested review from
adobrzyn,
afierka-intel,
iboiko-habana,
kamil-kaczor,
ksmusz,
mgawarkiewicz-intel and
michalkuligowski
as code owners
Contributor
There was a problem hiding this comment.
Pull request overview
This PR fixes a configuration issue in the Compressed Tensors implementation for HPU (Habana Processing Unit) to properly handle the fp8_inc KV cache dtype instead of the default fp8 format.
Changes:
- Added a custom
__init__method toHPUCompressedTensorsConfigthat overrides KV cache settings after parent initialization
💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.
xuechendi
approved these changes
Feb 5, 2026
✅ CI PassedAll checks passed successfully against the following vllm commit: |
adobrzyn
pushed a commit
that referenced
this pull request
Feb 6, 2026
Depends on #929 - Local test ```bash vllm ({'pretrained': '/mnt/disk5/hf_models/Qwen3-8B-FP8_STATIC-FP8-Attn-LLMC-Test-Only/', 'tensor_parallel_size': 8, 'max_model_len': 4096, 'max_num_seqs': 64, 'gpu_memory_utilization': 0.85, 'dtype': 'bfloat16', 'max_gen_toks': 2048, 'enable_prefix_caching': False, 'max_num_batched_tokens': 32768, 'kv_cache_dtype': 'fp8_inc'}), gen_kwargs: ({}), limit: None, num_fewshot: None, batch_size: 128 # |Tasks|Version| Filter |n-shot| Metric | |Value | |Stderr| # |-----|------:|----------------|-----:|-----------|---|-----:|---|-----:| # |gsm8k| 3|flexible-extract| 5|exact_match|↑ |0.8999|± |0.0083| # | | |strict-match | 5|exact_match|↑ |0.8999|± |0.0083| ``` cc @hshen14 @thuang6 --------- Signed-off-by: yiliu30 <yi4.liu@intel.com>
slokesha
pushed a commit
to libinta/vllm-gaudi
that referenced
this pull request
Feb 9, 2026
Adapt the update in vllm-project/vllm#30141 ```python # llm-compressor mdls need to set cache_dtype to "fp8" manually. if getattr(quant_config, "kv_cache_scheme", None) is not None: kv_cache_dtype = "fp8" calculate_kv_scales = False if cache_config is not None: cache_config.cache_dtype = "fp8" cache_config.calculate_kv_scales = False self.kv_cache_torch_dtype = kv_cache_dtype_str_to_dtype( kv_cache_dtype, vllm_config.model_config ) self.kv_cache_dtype = kv_cache_dtype ``` cc @hshen14 @thuang6 @lkk12014402 --------- Signed-off-by: yiliu30 <yi4.liu@intel.com> Signed-off-by: slokesha <slokeshappa@habana.ai>
slokesha
pushed a commit
to libinta/vllm-gaudi
that referenced
this pull request
Feb 9, 2026
Depends on vllm-project#929 - Local test ```bash vllm ({'pretrained': '/mnt/disk5/hf_models/Qwen3-8B-FP8_STATIC-FP8-Attn-LLMC-Test-Only/', 'tensor_parallel_size': 8, 'max_model_len': 4096, 'max_num_seqs': 64, 'gpu_memory_utilization': 0.85, 'dtype': 'bfloat16', 'max_gen_toks': 2048, 'enable_prefix_caching': False, 'max_num_batched_tokens': 32768, 'kv_cache_dtype': 'fp8_inc'}), gen_kwargs: ({}), limit: None, num_fewshot: None, batch_size: 128 # |Tasks|Version| Filter |n-shot| Metric | |Value | |Stderr| # |-----|------:|----------------|-----:|-----------|---|-----:|---|-----:| # |gsm8k| 3|flexible-extract| 5|exact_match|↑ |0.8999|± |0.0083| # | | |strict-match | 5|exact_match|↑ |0.8999|± |0.0083| ``` cc @hshen14 @thuang6 --------- Signed-off-by: yiliu30 <yi4.liu@intel.com> Signed-off-by: slokesha <slokeshappa@habana.ai>
adobrzyn
pushed a commit
that referenced
this pull request
Mar 31, 2026
Adapt the update in vllm-project/vllm#30141 ```python # llm-compressor mdls need to set cache_dtype to "fp8" manually. if getattr(quant_config, "kv_cache_scheme", None) is not None: kv_cache_dtype = "fp8" calculate_kv_scales = False if cache_config is not None: cache_config.cache_dtype = "fp8" cache_config.calculate_kv_scales = False self.kv_cache_torch_dtype = kv_cache_dtype_str_to_dtype( kv_cache_dtype, vllm_config.model_config ) self.kv_cache_dtype = kv_cache_dtype ``` cc @hshen14 @thuang6 @lkk12014402 --------- Signed-off-by: yiliu30 <yi4.liu@intel.com>
adobrzyn
pushed a commit
that referenced
this pull request
Mar 31, 2026
Depends on #929 - Local test ```bash vllm ({'pretrained': '/mnt/disk5/hf_models/Qwen3-8B-FP8_STATIC-FP8-Attn-LLMC-Test-Only/', 'tensor_parallel_size': 8, 'max_model_len': 4096, 'max_num_seqs': 64, 'gpu_memory_utilization': 0.85, 'dtype': 'bfloat16', 'max_gen_toks': 2048, 'enable_prefix_caching': False, 'max_num_batched_tokens': 32768, 'kv_cache_dtype': 'fp8_inc'}), gen_kwargs: ({}), limit: None, num_fewshot: None, batch_size: 128 # |Tasks|Version| Filter |n-shot| Metric | |Value | |Stderr| # |-----|------:|----------------|-----:|-----------|---|-----:|---|-----:| # |gsm8k| 3|flexible-extract| 5|exact_match|↑ |0.8999|± |0.0083| # | | |strict-match | 5|exact_match|↑ |0.8999|± |0.0083| ``` cc @hshen14 @thuang6 --------- Signed-off-by: yiliu30 <yi4.liu@intel.com>
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
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.
Adapt the update in vllm-project/vllm#30141
cc @hshen14 @thuang6 @lkk12014402