minimind

Commit Graph

Author	SHA1	Message	Date
baihaochen	33233fed6e	feat: add SearchShortQA DeepSpeed pipeline	2026-06-02 00:59:46 +08:00
baihaochen	bc3d137248	[feat] add MLA with decoupled RoPE (DeepSeek-V2 style) - Add model_minimind_mla.py with Q/KV low-rank compression and decoupled RoPE - Add kv_norm on kv_latent, q_compress + q_up + q_rope_proj for Q path - Cache format: (kv_latent, k_rope) tuple, 176 floats/tok/layer (4.4x compression) - Add --use_mla flag to all training scripts and utils - Add benchmark_gqa_vs_mla.py script	2026-05-12 18:38:47 +08:00

Author

SHA1

Message

Date

baihaochen

33233fed6e

feat: add SearchShortQA DeepSpeed pipeline

2026-06-02 00:59:46 +08:00

baihaochen

bc3d137248

[feat] add MLA with decoupled RoPE (DeepSeek-V2 style)

- Add model_minimind_mla.py with Q/KV low-rank compression and decoupled RoPE
- Add kv_norm on kv_latent, q_compress + q_up + q_rope_proj for Q path
- Cache format: (kv_latent, k_rope) tuple, 176 floats/tok/layer (4.4x compression)
- Add --use_mla flag to all training scripts and utils
- Add benchmark_gqa_vs_mla.py script

2026-05-12 18:38:47 +08:00

2 Commits (5012b232ff535d5c9917ef2f06eab38927cc0a3a)