Skip to content

Pull requests: neuralmagic/compressed-tensors

New pull request New
24 Open 356 Closed
24 Open 356 Closed
Author
Filter by author
Loading
Label
Filter by label
Loading
Use alt + click/return to exclude labels
or ⇧ + click/return for logical OR
Projects
Filter by project
Loading
Milestones
Filter by milestone
Loading
Reviews
Filter by reviews
No reviews Review required Approved review Changes requested
Assignee
Filter by who’s assigned
Assigned to nobody Loading
Sort
Sort by
Newest Oldest Most commented Least commented Recently updated Least recently updated Best match
Most reactions
👍 👎 😄 🎉 😕 ❤️ 🚀 👀

Pull requests list

[Quantization][Decompression] Fix QDQ for dynamic quant; Update NVFP4 Compression Params
#407 opened Jul 28, 2025 by dsikka • Draft
Refactor module / parameter matching logic
#406 opened Jul 28, 2025 by fynnsu Loading…
4 of 6 tasks
21
Raise ValueError when nvfp4 pack tensor has odd number of columns
#402 opened Jul 23, 2025 by fynnsu Loading…
1
4
Speed up nvfp4 pack/unpack w/ torch.compile
#400 opened Jul 22, 2025 by fynnsu Loading…
6
[KV Cache] support kv cache int8 per channel quant
#398 opened Jul 19, 2025 by Eviannn Loading…
[Transform] R3 and R4 attention rotations
#397 opened Jul 16, 2025 by kylesayrs • Draft
[Bugfix] Fix Per-Token Dynamic Activation Quantization
#393 opened Jul 14, 2025 by max410011 Loading…
5
[Transform] [WIP] Transforms HEAD
#391 opened Jul 11, 2025 by kylesayrs • Draft
2
Error when configs are created with unrecognized fields
#386 opened Jul 10, 2025 by kylesayrs Loading…
1
[WIP] [Research] Attention quantization and transformation
#374 opened Jul 1, 2025 by kylesayrs • Draft
[Transform] Serialize with tied weights
#370 opened Jun 28, 2025 by kylesayrs Loading…
5
[WIP] Initialize qparams with valid values
#368 opened Jun 23, 2025 by kylesayrs • Draft
[Accelerate] Remove is_module_offloaded and update_prefix_dict
#366 opened Jun 20, 2025 by kylesayrs • Draft
[Accelerate] Support get_offloaded_device for models
#364 opened Jun 20, 2025 by kylesayrs • Draft
[Bugfix] Fix updating parameters under align_modules
#361 opened Jun 18, 2025 by kylesayrs Loading…
Optimize sparse 2:4 compression performance
#358 opened Jun 16, 2025 by rahul-tuli • Draft
8 tasks done
relax setuptools_scm version requirement
#343 opened Jun 6, 2025 by envolution Loading…
[Misc] Remove unused config name definitions
#332 opened May 31, 2025 by kylesayrs Loading…
2
Make MSE the default
#324 opened May 22, 2025 by dsikka Loading…
4
[Transforms][WIP] Update wrapped_forward to use hooks; apply transforms to activations
#286 opened Mar 31, 2025 by dsikka • Draft
1
[Transforms] Initial Implementation
#277 opened Mar 11, 2025 by dsikka • Draft
[Transforms] Apply, serialize, deserialize
#276 opened Mar 11, 2025 by dsikka • Draft
6
[Quantization] Channel wise output activation quantization for QKV Attention layers
#270 opened Mar 7, 2025 by horheynm • Draft
Harden KV Cache qparams
#262 opened Feb 26, 2025 by kylesayrs • Draft
3
ProTip! Add no:assignee to see everything that’s not assigned.