Automatically tell users that dict args must be valid JSON in CLI (#17577)

Harry Mellor committed over 1 year ago

[doc] add the print result (#17584)

Reid committed over 1 year ago

add more pytorch related tests for torch nightly (#17422)

Yang Wang committed over 1 year ago

[Misc] Rename assets for testing (#17575)

Cyrus Leung committed over 1 year ago

[BugFix] Fix Memory Leak (#17567)

Robert Shaw committed over 1 year ago

[Core] [Bugfix] Add Input Embeddings (#15428)

Andrew Sansom committed over 1 year ago

[Bugifx] Remove TritonPlaceholder from sys.modules (#17317)

Isotr0py committed over 1 year ago

Add `pt_load_map_location` to allow loading to cuda (#16869)

Jerry Zhang committed over 1 year ago

Fix PixtralHF missing spatial_merge_size (#17571)

Michael Goin committed over 1 year ago

[Misc] Continue refactoring model tests (#17573)

Cyrus Leung committed over 1 year ago

Check if bitblas is installed during support check (#17572)

Michael Goin committed over 1 year ago

[CI] Actually run tests/kv_transfer/test_disagg.py in CI (#17555)

Michael Goin committed over 1 year ago

[Attention] MLA move o_proj q_proj into cuda-graph region (#17484)

Lucas Wilkinson committed over 1 year ago

[Doc] note that not all unit tests pass on CPU platforms (#17554)

David Xia committed over 1 year ago

[Bugfix] Disable gptq_bitblas for <SM80 to fix GPTQ on V100/T4 (#17541)

Michael Goin committed over 1 year ago

[V1][Spec Decode] Apply torch.compile & cudagraph to EAGLE3 (#17504)

qizixi committed over 1 year ago

[Misc][Tools][Benchmark] Publish script to auto tune server parameters (#17207)

Chenyaaang committed over 1 year ago

[Bug]change the position of cuda_graph_sizes in dataclasses (#17548)

Chen Xia committed over 1 year ago

Remove duplicate code from dbrx.py (#17550)

sstamenk committed over 1 year ago

[ROCM] Add gfx950 to the custom attention archs (#16034)

Juan Villamizar committed over 1 year ago