67a5639c3c
Based on a3c45d3b (last known working docker build):
- yaml: max-num-seqs=2 (fixes t2_n_2), TOPK_SOFTMAX=1 (use prebuilt .so)
- yaml: keep max-model-len=131072, gpu-mem=0.90 (prevents OOM)
- yaml: NO LD_PRELOAD (libcccl not built during docker build)
- xformers: revert to Q-tiling only (flash_attn caused OOM at profiling)
- .dockerignore: exclude all non-essential files from context
- remove libcccl_allocator.so from git tracking
What stays from recent work:
- 14 prebuilt .so (including corex_gdn_chunk_recurrent)
- qwen3_5.py with .float() fix and chunk_recurrent support
- All vendor_overrides and CCCL preload source (for future use)
42 lines
597 B
Plaintext
42 lines
597 B
Plaintext
**/__pycache__
|
|
**/*.pyc
|
|
**/.git
|
|
cccl_upstream/
|
|
upstream_ref/
|
|
vllm/
|
|
ixformer_sdk/
|
|
muh/
|
|
ex_engine/
|
|
*.zip
|
|
enginex-vllm-bi100-qwen36-main.zip
|
|
dockerrizhi.txt
|
|
subrizhi.txt
|
|
docs/
|
|
optimizations/
|
|
vllm_adapter/
|
|
vllm_overrides/
|
|
*.md
|
|
verify_*.py
|
|
verify_*.sh
|
|
test_*.py
|
|
debug_*.py
|
|
probe_*.py
|
|
probe_*.sh
|
|
diagnose_*.sh
|
|
cat_*.py
|
|
attention.py
|
|
paged_attention_v2_*.py
|
|
paged_attn.py
|
|
prefix_prefill.py
|
|
engine_cccl_patterns.py
|
|
deltanet_chunk_optimize.py
|
|
muh_*.py
|
|
__init__.py
|
|
chat_dataset_v0.json
|
|
cccl_sm100_benchmark_values.json
|
|
baseline.muh
|
|
launch_service
|
|
Dockerfile.*
|
|
computility-run.*.yaml
|
|
computility-run.yaml.bak
|