Michelle0574
4f84738c0f
Merge branch 'feat/moe-layer' of github.com:RYDE-WORK/lnp_ml into feat/moe-layer
2026-08-15 12:31:54 +00:00
Michelle0574
aeddf8e1ee
feat: 支持固定超参训练并部署 4 专家稀疏 MoE 模型
2026-08-15 12:22:21 +00:00
Dicong Li
67db7fa473
Add pretrained encoder baseline scripts and results
2026-08-14 17:37:31 +00:00
Michelle0574
c0203ebe76
exp: 5折固定超参对比实验
2026-08-14 16:59:58 +00:00
Michelle0574
866941039f
feat: final_train_optuna_cv 支持 MoE + Qwen2.5-7B QLoRA + soft-RAG，产出可部署 checkpoint
2026-08-13 10:05:00 +00:00
Michelle0574
bab8203c98
Add MoleculeSTM/MolE/RF/KNN baselines: encoders, scripts, and CV results
2026-07-28 14:14:10 +00:00
Michelle0574
39a2a98fd9
Add full ablation results and CheMeleon/UniMol encoders
2026-07-24 17:56:49 +00:00
Michelle0574
22e83122e7
feat: Qwen SoftRAG（RAG prompt/soft token/QLoRA）+ Morgan 检索旁路与实验结果
2026-07-07 17:47:34 +00:00
Michelle0474
c0aa30f2bb
Add 6-variant ablation results (baseline/moe/llm_molt5/llm_biot5/both_molt5/both_biot5)
...
- Updated ablation_summary.csv with full 6-variant comparison
- Updated run_all.sh to run 6 variants (added BioT5 variants)
- Added per-fold json artifacts (best_params/test_metrics/splits/history/summary)
- Model weights (.pt) excluded per .gitignore
2026-06-24 15:33:07 +08:00
Michelle0574
cd9166d63c
feat(moe): MoE + MolT5 LLM 融合层、消融流水线
2026-06-18 08:51:30 +00:00
