fix(mmlu): expert-DP batch sharding for MoE + Nano-30B-A3B PTQ example#1967

Merged

ChenhanYu merged 13 commits into

mainfrom

chenhany/repro-5434-validation

Jul 14, 2026

Commits on Jul 13, 2026

examples(llama-qad): lower MMLU gate 0.40 -> 0.36 for NVFP4 Llama-3.2-1B

ChenhanYu
and
claude
committed
Merge branch 'main' into chenhany/repro-5434-validation
ChenhanYu
authored

Commits on Jul 14, 2026

test(megatron): skip flaky compressed sharded_state_dict on Blackwell

ChenhanYu
and
claude
committed
example(eagle3): Qwen3-0.6B EAGLE3 online/offline launcher examples

ChenhanYu
and
claude
committed
example(eagle3): fix Qwen3-0.6B offline parallelism + cap query samples
ChenhanYu
committed
example(eagle3): Qwen3-0.6B training_seq_len 4096 -> 1024
ChenhanYu
committed
example(eagle3): disable torch.compile for Qwen3-0.6B CI train
ChenhanYu
committed
fix(dataset): retry HF datasets-server /splits; xfail flaky preprocess tests
ChenhanYu
committed
Merge branch 'main' into chenhany/repro-5434-validation
ChenhanYu
committed