Skip to content

Actions: gaugarg-nv/llama.cpp

Actions

Python Type-Check

Actions

Loading...
Loading

Show workflow options

Create status badge

Loading
python-type-check.yml
23 workflow runs
23 workflow runs

Filter by Event

⌥ + click/return to exclude

Filter by Status

⌥ + click/return to exclude

Filter by Branch

⌥ + click/return to exclude

Filter by Actor

⌥ + click/return to exclude
Enable Auto parameter fitting in split-mode tensor Python Type-Check #25: Commit f46c497 pushed by gaugarg-nv
3m 22s tp_fit_support
tp_fit_support
3m 22s
Python Type-Check Python Type-Check #24: by gaugarg-nv
3m 38s b9037
b9037
3m 38s
Write a readme on Multi-GPU usage in llama.cpp Python Type-Check #23: Commit f3f02e1 pushed by gaugarg-nv
4m 27s multi-gpu-read-me
multi-gpu-read-me
4m 27s
Fix recurrent state serialization for partial reads and writes Python Type-Check #22: Commit 92f0041 pushed by gaugarg-nv
3m 31s fix_cache_serialization
fix_cache_serialization
3m 31s
Fix delayed AllReduce on Gemma-4 MoE Python Type-Check #21: Commit 4ce8fde pushed by gaugarg-nv
3m 13s gemma4_perf
gemma4_perf
3m 13s
cache subgraph splits when cgraph is unchanged Python Type-Check #20: Commit 4bfe714 pushed by gaugarg-nv
2m 39s tp_cpu_overhead
tp_cpu_overhead
2m 39s
Change CMake to use NCCL on Windows Python Type-Check #19: Commit 68fb0ba pushed by gaugarg-nv
2m 33s nccl_windows
nccl_windows
2m 33s
Change CMake to use NCCL on Windows Python Type-Check #18: Commit a92f245 pushed by gaugarg-nv
3m 23s nccl_windows
nccl_windows
3m 23s
Use the new kernel only for nblocks_stream_k_raw > 4 * ntiles_dst to … Python Type-Check #17: Commit 25ef2df pushed by gaugarg-nv
2m 36s fa_opt
fa_opt
2m 36s
Change CMake to use NCCL on Windows Python Type-Check #16: Commit 23e1600 pushed by gaugarg-nv
3m 4s nccl_windows
nccl_windows
3m 4s
Write an optimized flash_attn_stream_k_fixup kernel Python Type-Check #15: Commit 99c3df8 pushed by gaugarg-nv
3m 4s fa_opt
fa_opt
3m 4s
Reduce the number of stream-k blocks to reduce the overhead of the fl… Python Type-Check #14: Commit 38a6935 pushed by gaugarg-nv
2m 41s reduce_stream_k_block
reduce_stream_k_block
2m 41s
Cherry-pick changes from @am17an PR https://github.com/ggml-org/llama… Python Type-Check #13: Commit 5d282ad pushed by gaugarg-nv
2m 25s moe_opt
moe_opt
2m 25s
Optimize MOE GEMV kernel for BS > 1. Python Type-Check #12: Commit 7cfb67e pushed by gaugarg-nv
3m 25s moe_opt
moe_opt
3m 25s
Increase per-thread work if the K-dimension is small Python Type-Check #11: Commit cfbbfb2 pushed by gaugarg-nv
2m 2s small_k_optimization
small_k_optimization
2m 2s
Use a single warp per element instead of a single block per element i… Python Type-Check #10: Commit 4f20a44 pushed by gaugarg-nv
2m 19s small_k_optimization
small_k_optimization
2m 19s
Use a single warp per element instead of a single block per element i… Python Type-Check #9: Commit 2afd6a3 pushed by gaugarg-nv
2m 5s small_k_optimization
small_k_optimization
2m 5s
Change CMake to use NCCL on Windows Python Type-Check #8: Commit 6c90394 pushed by gaugarg-nv
1m 55s nccl_windows
nccl_windows
1m 55s
Improve CUDA graph capture Python Type-Check #7: Commit a874a2c pushed by gaugarg-nv
1m 58s cuda_graph_fix
cuda_graph_fix
1m 58s
Fix the seg fault without NCCL Python Type-Check #6: Commit 10385e8 pushed by gaugarg-nv
1m 54s minor_fixes
minor_fixes
1m 54s
Fix Issue !19219 Python Type-Check #5: Commit 027ef82 pushed by gaugarg-nv
2m 21s revert_scale_queue_env
revert_scale_queue_env
2m 21s
[CUDA] Reduce CPU-side stalls due to the CUDA command buffer being full Python Type-Check #4: Commit d3298dc pushed by gaugarg-nv
2m 4s pp_perf_improve
pp_perf_improve
2m 4s
Prefer vector flash decoding kernel for Gemma models Python Type-Check #3: Commit f7d07dd pushed by gaugarg-nv
1m 29s gemma_flash_attention
gemma_flash_attention
1m 29s