 Sigbjørn Skjæret
|
e98b3692be
llama : set qwen3 model type sizes (#13175)
|
пре 8 месеци |
 AT
|
5f5e39e1ba
model : Nomic Embed Text V2 with Mixture-of-Experts (MoE) architecture (#12466)
|
пре 8 месеци |
 Johannes Gäßler
|
69699be48a
CUDA: fix q_nope_absorbed prec for DS 2 Lite f16 (#13137)
|
пре 9 месеци |
 Georgi Gerganov
|
2f74c354c0
graph : make FA compatible with MLA + add initial Metal kernels (#12953)
|
пре 9 месеци |
 Juk Armstrong
|
daa422881a
llama : DeepSeek V2/V3 MLA implementation (#12801)
|
пре 9 месеци |
 Yuxuan Zhang
|
06bb53ad9b
llama-model : add Glm4Model implementation for GLM-4-0414 (#12867)
|
пре 9 месеци |
 Xuan-Son Nguyen
|
8b91d5355a
llama : correct rms norm for llama 4 (#12882)
|
пре 9 месеци |
 Bo Zheng
|
d3bd7193ba
llama : Support Qwen3 and Qwen3MoE (#12828)
|
пре 9 месеци |
|
Xuan-Son Nguyen
|
1466621e73
llama : Support llama 4 text-only (#12791)
|
пре 9 месеци |
 Diego Devesa
|
e0e912f49b
llama : add option to override model tensor buffers (#11397)
|
пре 9 месеци |
|
Sigbjørn Skjæret
|
2c3f8b850a
llama : support BailingMoE (Ling) (#12634)
|
пре 9 месеци |
 Djip007
|
0bb2919335
llama : change cpu_buft_list order: ACCEL -> GPU host -> CPU extra -> CPU (#12632)
|
пре 9 месеци |
|
Sigbjørn Skjæret
|
3714c3ee1a
llama : fix incorrect Qwen2Moe ffn_moe_out graph callback (#12631)
|
пре 10 месеци |
 Si1w
|
f125b8dccf
llama : add PLM GGUF Conversion & Inference Support (#12457)
|
пре 10 месеци |
 HighDoping
|
953c2a62cf
model : restore support for T5Encoder (#12590)
|
пре 10 месеци |
|
Xuan-Son Nguyen
|
fbdfefe74e
llama : gemma3 : use output tensor if it exists in model weight (#12506)
|
пре 10 месеци |
|
Georgi Gerganov
|
af04481e6b
model : do not repack if a GPU device is present (#12498)
|
пре 10 месеци |
|
Sigbjørn Skjæret
|
960e726077
chore : cleanup llama_model_loader::TENSOR_ usage (#12492)
|
пре 10 месеци |
|
Sigbjørn Skjæret
|
dbb3a4739e
llama : make Qwen2MoE QKV bias optional (#12477)
|
пре 10 месеци |
|
Sigbjørn Skjæret
|
108e53c2f1
llama : add support for GPT2, Bloom and CodeShell tied word embeddings (#12456)
|
пре 10 месеци |
|
Georgi Gerganov
|
75422e8bc4
graph : normalize Q, K, V shapes + sync cross attention (#12449)
|
пре 10 месеци |
|
Xuan-Son Nguyen
|
99aa304fb9
llama : add support for EXAONE tied word embeddings (#12451)
|
пре 10 месеци |
 Molly Sophia
|
7dfad387e3
llama: Add support for RWKV v7 architecture (#12412)
|
пре 10 месеци |
|
Sigbjørn Skjæret
|
8ba95dca20
llama : fix OLMo-2-0325-32B-Instruct K-norm size (#12400)
|
пре 10 месеци |
|
Georgi Gerganov
|
c522ce4143
graph : simplify attn input build for unified KV cache (#12381)
|
пре 10 месеци |
|
Georgi Gerganov
|
081bee8c64
hparams : add SWA rope parameters (#12374)
|
пре 10 месеци |
|
Georgi Gerganov
|
84d5475541
llama : fix Gemma3 SWA KV cache shift (#12373)
|
пре 10 месеци |
|
Georgi Gerganov
|
e0dbec0bc6
llama : refactor llama_context, llama_kv_cache, llm_build_context (#12181)
|
пре 10 месеци |
|
Xuan-Son Nguyen
|
7841fc723e
llama : Add Gemma 3 support (+ experimental vision capability) (#12343)
|
пре 10 месеци |
 Xuan-Son Nguyen
|
c43a3e7996
llama : add Phi-4-mini support (supersede #12099) (#12108)
|
пре 10 месеци |
