Michael Yang adff143bcd fix: mllama quality (#10807) ...

* fix mllama convert

- transform attn_gate and ffn_gate
- swap attention heads for vision models

* fix mllama

the mlp gate which was applied in the wrong place

2025-05-22 11:30:49 -07:00

..

gemma2

ml: add more rope options (#10775)

2025-05-20 15:51:08 -07:00

gemma3

ml: add more rope options (#10775)

2025-05-20 15:51:08 -07:00

llama

feat: port qwen2 model (#10782)

2025-05-21 10:21:24 -07:00

llama4

feat: qwen3 dense and sparse models (#10708)

2025-05-21 10:21:07 -07:00

mistral3

ml: add more rope options (#10775)

2025-05-20 15:51:08 -07:00

mllama

fix: mllama quality (#10807)

2025-05-22 11:30:49 -07:00

qwen2

feat: port qwen2 model (#10782)

2025-05-21 10:21:24 -07:00

qwen3

feat: qwen3 dense and sparse models (#10708)

2025-05-21 10:21:07 -07:00

qwen25vl

fix: qwen25vl assign samebatch in multimodal input (#10789)

2025-05-21 09:39:20 -07:00

models.go

feat: port qwen2 model (#10782)

2025-05-21 10:21:24 -07:00