miles

amd

[AMD] DeepSeek-V4: use the ROCm precision-parity norm path (#2040 )

2026-07-31 15:09:50 -07:00

models

[Lora] Add GDN (Qwen3.5/Qwen3-Next) LoRA target-module mapping (#1597 )

2026-07-15 14:15:35 -07:00

tools

refactor(tito): remove --tito-allowed-append-roles and derive append surfaces from fixed templates (#1818 )

2026-07-27 13:45:04 -07:00

run_deepseek_v4.py

refactor(examples): group examples into infra_features/ and experimental/ (#1953 )

2026-07-29 12:10:19 -07:00

run_deepseek_v32.py

refactor(examples): group examples into infra_features/ and experimental/ (#1953 )

2026-07-29 12:10:19 -07:00

run_deepseek.py

chore: replace huggingface-cli with hf (#879 )

2026-04-04 10:08:14 +08:00

run_gemma_4_26b_a4b.py

Remove deprecated --use-miles-router flag from launch scripts (#1480 )

2026-06-26 12:02:47 +09:00

run_gemma_4_31b.py

[feat] Gemma-4 model support (#1232 )

2026-06-24 12:44:14 -07:00

run_glm5_1_744b_a40b_lora.py

fix(ci): align model-script jobs with GPU allocation (#1671 )

2026-07-14 17:35:50 -07:00

run_glm5_2_744b_a40b_lora.py

[WIP] feat: lora GLM-5.2 FP8 rollout support (#1678 )

2026-07-15 16:22:36 -07:00

run_glm5_2_744b_a40b.py

fix GLM 5 & 5.2 on blackwell (#1477 )

2026-06-25 18:52:49 -07:00

run_glm5_744b_a40b.py

fix GLM 5 & 5.2 on blackwell (#1477 )

2026-06-25 18:52:49 -07:00

run_glm45_355b_a32b.py

refactor(examples): group examples into infra_features/ and experimental/ (#1953 )

2026-07-29 12:10:19 -07:00

run_glm47_flash.py

fix glm47-flash: use paged MLA prefill on B200 (#2047 )

2026-08-02 19:38:11 -07:00

run_joy_ai_llm_flash.py

refactor(examples): group examples into infra_features/ and experimental/ (#1953 )

2026-07-29 12:10:19 -07:00

run_kimi_k25.py

fix(ci): align model-script jobs with GPU allocation (#1671 )

2026-07-14 17:35:50 -07:00

run_mcore_fsdp.py

fix: pass hf_checkpoint to convert_checkpoint using model_dir (#749 )

2026-03-18 16:12:50 +08:00

run_nemotron_3_nano_4b_fsdp.py

[fsdp] fsdp backend support fix (#1469 )

2026-07-27 17:02:43 -07:00

run_qwen3_0_6b_fsdp.py

[fsdp] fsdp backend support fix (#1469 )

2026-07-27 17:02:43 -07:00

run_qwen3_4b_npu.py

Fix pre-commit formatting (#1263 )

2026-05-29 16:05:21 -07:00

run_qwen3_4b_npu.sh

npu support qwen3-4b. (#1125 )

2026-05-29 17:18:16 +08:00

run_qwen3_4b.py

refactor(examples): group examples into infra_features/ and experimental/ (#1953 )

2026-07-29 12:10:19 -07:00

run_qwen3_5_35b_a3b_lora.py

[Lora] Add GDN (Qwen3.5/Qwen3-Next) LoRA target-module mapping (#1597 )

2026-07-15 14:15:35 -07:00

run_qwen3_5_35b_a3b_mtp_cp2_ep8.py

Megatron e2e: weight-check skip-list, Qwen3.5 MTP cases (#1512 )

2026-06-30 15:05:11 -07:00

run_qwen3_6_35b_a3b_mtp.py

Feat: Qwen3.6 RL support (#1022 )

2026-04-30 17:57:03 -07:00

run_qwen3_30b_a3b_fsdp.py

[fsdp] fsdp backend support fix (#1469 )

2026-07-27 17:02:43 -07:00

run_qwen3_30b_a3b.py

refactor(examples): group examples into infra_features/ and experimental/ (#1953 )

2026-07-29 12:10:19 -07:00

run-deepseek-r1.sh

Fix example rollout_temperature and top_k (#338 )

2025-12-23 16:30:15 -08:00

run-glm4-9B-4xgpu-radixtree.sh

Remove deprecated --use-miles-router flag from launch scripts (#1480 )

2026-06-26 12:02:47 +09:00

run-glm4-9B.sh

Fix example rollout_temperature and top_k (#338 )

2025-12-23 16:30:15 -08:00

run-glm4.5-355B-A32B.sh

Fix example rollout_temperature and top_k (#338 )

2025-12-23 16:30:15 -08:00

run-glm4.7-flash.sh

Remove deprecated --use-miles-router flag from launch scripts (#1480 )

2026-06-26 12:02:47 +09:00

run-gpt-oss-20b-bf16.sh

[model] Add gpt-oss-20b-bf16 script (#590 )

2026-03-12 15:45:19 -07:00

run-kimi-k2-Instruct.sh

Fix example rollout_temperature and top_k (#338 )

2025-12-23 16:30:15 -08:00

run-kimi-k2-Thinking.sh

Fix example rollout_temperature and top_k (#338 )

2025-12-23 16:30:15 -08:00

run-kimi-k25.sh

Remove deprecated --use-miles-router flag from launch scripts (#1480 )

2026-06-26 12:02:47 +09:00

run-mimo-7B-rl-eagle.sh

Fix example rollout_temperature and top_k (#338 )

2025-12-23 16:30:15 -08:00

run-moonlight-16B-A3B.sh

Fix example rollout_temperature and top_k (#338 )

2025-12-23 16:30:15 -08:00

run-nemotron-3-nano-4b.sh

Feat: NVIDIA Nemotron-3 RL support (#1033 )

2026-04-30 17:57:24 -07:00

run-nemotron-3-nano-30b-a3b.sh

Remove deprecated --use-miles-router flag from launch scripts (#1480 )

2026-06-26 12:02:47 +09:00

run-nemotron-3-super-120b-a12b.sh

Remove deprecated --use-miles-router flag from launch scripts (#1480 )

2026-06-26 12:02:47 +09:00

run-qwen3-4B_4xgpu.sh

Remove deprecated --use-miles-router flag from launch scripts (#1480 )

2026-06-26 12:02:47 +09:00

run-qwen3-4B-base-sft.sh

update code (#385 )

2026-01-03 20:09:56 -08:00

run-qwen3-4B.sh

Feat/add prometheus logging (#757 )

2026-03-24 09:42:55 +08:00

run-qwen3-32B.sh

Fix example rollout_temperature and top_k (#338 )

2025-12-23 16:30:15 -08:00

run-qwen3-235B-A22B-sft.sh

update code (#401 )

2026-01-05 13:32:45 -08:00

run-qwen3-235B-A22B.sh

fix: missing comma in runtime env JSON for qwen3-235B-A22B (#531 )

2026-01-30 18:45:12 +08:00

run-qwen3-next-80B-A3B-8gpus.sh

[auto-sync] update code (#383 )

2026-01-02 18:42:42 -08:00

run-qwen3-next-80B-A3B.sh

Fix example rollout_temperature and top_k (#338 )

2025-12-23 16:30:15 -08:00

run-qwen3.5-4B.sh

[model] Add Qwen3.5 model support (4B, 9B, 27B and 35B-A3B) (#740 )

2026-03-23 16:16:27 -07:00

run-qwen3.5-9B.sh

[model] Add Qwen3.5 model support (4B, 9B, 27B and 35B-A3B) (#740 )

2026-03-23 16:16:27 -07:00

run-qwen3.5-27B.sh

[model] Add Qwen3.5 model support (4B, 9B, 27B and 35B-A3B) (#740 )

2026-03-23 16:16:27 -07:00

run-qwen3.5-35B-A3B-mtp.sh

[model] Add Qwen3.5 model support (4B, 9B, 27B and 35B-A3B) (#740 )

2026-03-23 16:16:27 -07:00

run-qwen3.6-27B.sh

Feat: Qwen3.6 RL support (#1022 )

2026-04-30 17:57:03 -07:00