Repository navigation
[None][feat] Qwen3.8-Flash-Next NVFP4 modeling_v2 unified target on 4×GB300 (TP4/TEP4/ADP4), AgentX-tuned - #21372
Draft
Wanli-Jiang wants to merge 218 commits into
Draft
Wanli-Jiang wants to merge 218 commits into
Wanli-Jiang wants to merge 218 commits into
Commits
Commits on Oct 30, 2026
- committed
- committed
- committed
- committed
[TRTLLM-16460][perf] Qwen3.8-Flash-Next NVFP4 target: GB300 Pareto optimizations for ISL 8K / OSL 1K
committed[TRTLLM-16460][test] Qwen3.8-Flash-Next GB300 perf sanity on the target's fastest MTP3 configuration
committed- committed
Commits on Oct 31, 2026
- committed
- committed
- committed
- committed
- committed
- committed
- committed
- committed
- committed
- committed
- committed
- committed
- committed
- committed
- committed
- committed
- committed
- committed
- committed
- committed
- committed
- committed
- committed
- committed
- committed
- committed
[None][feat] modeling_v2 qwen4_exp unified: ple_ngram_embed__r15 variant, copied from the tp1 target
committed- committed
- committed
- committed
- committed
- committed
- committed
- committed
- committed
- committed
- committed
- committed
- committed
- committed
- committed
- committed
[None][feat] modeling_v2 qwen4_exp unified: qsa_topk_select__r23 variant, copied from the tp1 target
committed- committed
- committed
- committed
- committed
- committed
- committed
- committed
- committed
- committed
- committed
- committed
- committed
- committed
- committed
- committed
- committed
- committed
- committed
- committed
- committed
- committed
- committed
- committed
- committed
- committed
- committed
- committed
- committed
- committed
- committed
- committed
- committed
- committed
- committed
- committed
- committed
- committed
- committed
- committed
- committed
- committed
- committed
- committed
- committed
- committed
- committed
- committed
- committed
- committed
- committed
- committed
- committed
- committed
- committed
- committed
- committed
- committed
- committed
- committed
- committed
- committed
- committed
- committed
- committed
- committed
- committed
- committed
- committed
- committed
- committed
- committed
- committed
- committed
- committed
- committed
[None][test] modeling_v2 qwen4_exp unified: the tep4 core's per-rank widths and expert window on CPU
committed- committed
- committed
- committed
- committed
- committed
- committed
- committed
- committed
- committed
- committed
- committed
- committed
- committed
- committed
- committed
- committed
- committed
- committed
- committed
- committed
- committed
- committed
- committed
- committed
- committed
- committed
- committed
- committed
- committed
- committed
- committed
[None][chore] modeling_v2 catalog: activation/silu_and_mul_rows, promoted from the qwen4_exp targets
committed- committed
- committed
- committed
[None][chore] modeling_v2 catalog: attention/qsa_mtp_reuse_topk, promoted from the qwen4_exp targets
committed- committed
- committed
[None][chore] modeling_v2 catalog: attention/qsa_paged_kv_store, promoted from the qwen4_exp targets
committed- committed
- committed
- committed
- committed
- committed
- committed
- committed
- committed
- committed
- committed
- committed
- committed
- committed
[None][chore] modeling_v2 catalog: hc/qwen4_exp_hc_up_gate_mean, promoted from the qwen4_exp targets
committed- committed
- committed
- committed
- committed
- committed
- committed
- committed
- committed
- committed
- committed
- committed
- committed
- committed
[None][chore] modeling_v2 catalog: ssm/cu_seqlens_from_seq_lens, promoted from the qwen4_exp targets
committed- committed
- committed
- committed
- committed
- committed
- committed
- committed
- committed
- committed
- committed
- committed
- committed
- committed
- committed
- committed
- committed
- committed
- committed
- committed
- committed
- committed
- committed
- committed
Commits on Nov 1, 2026
