nikola/llama_cpp
1
0
Fork 0
Code Issues Pull Requests Actions 17 Packages Projects Releases Wiki Activity

model: support youtu-vl model (#18479)

Browse Source 
* Support Youtu-VL Model

* merge code

* fix bug

* revert qwen2 code & support rsplit in minja.hpp

* update warm info

* fix annotation

* u

* revert minja.hpp

* fix

* Do not write routed_scaling_factor to gguf when routed_scaling_factor is None

* fix expert_weights_scale

* LGTM after whitespace fixes

* fix

* fix

* fix

* layers to layer_index

* enum fix

---------

Co-authored-by: Xuan-Son Nguyen <son@huggingface.co>
Co-authored-by: Sigbjørn Skjæret <sigbjorn.skjaeret@scala.com>
This commit is contained in:
tt
2026-01-01 19:25:54 +01:00
committed by GitHub
co-authored by Xuan-Son Nguyen Sigbjørn Skjæret
parent 3ccccc83f7
commit ced765be44
17 changed files with 473 additions and 36 deletions
Download Patch File Download Diff File Expand all files Collapse all files
gguf-py/gguf/tensor_mapping.py
+12
View File
@@ -1221,6 +1221,7 @@ class TensorNameMap:
MODEL_TENSOR.V_MMPROJ: (
"multi_modal_projector.linear_{bid}",
"visual.merger.mlp.{bid}", # qwen2vl
"merger.mlp.{bid}",
),
MODEL_TENSOR.V_MMPROJ_FC: (
@@ -1258,6 +1259,7 @@ class TensorNameMap:
"visual.patch_embed.proj", # qwen2vl
"vision_tower.patch_embed.proj", # kimi-vl
"model.vision.patch_embedding.proj", # cogvlm
"siglip2.vision_model.embeddings.patch_embedding",
),
MODEL_TENSOR.V_ENC_EMBD_NORM: (
@@ -1291,6 +1293,7 @@ class TensorNameMap:
"vision_encoder.transformer.layers.{bid}.attention.wq", # pixtral
"visual.blocks.{bid}.attn.q", # qwen2vl, generated
"vision_tower.encoder.blocks.{bid}.wq", # kimi-vl, generated
"siglip2.vision_model.encoder.layers.{bid}.self_attn.q_proj", # youtuvl
),
MODEL_TENSOR.V_ENC_ATTN_Q_NORM: (
@@ -1308,6 +1311,7 @@ class TensorNameMap:
"vision_encoder.transformer.layers.{bid}.attention.wk", # pixtral
"visual.blocks.{bid}.attn.k", # qwen2vl, generated
"vision_tower.encoder.blocks.{bid}.wk", # kimi-vl, generated
"siglip2.vision_model.encoder.layers.{bid}.self_attn.k_proj",
),
MODEL_TENSOR.V_ENC_ATTN_K_NORM: (
@@ -1325,6 +1329,7 @@ class TensorNameMap:
"vision_encoder.transformer.layers.{bid}.attention.wv", # pixtral
"visual.blocks.{bid}.attn.v", # qwen2vl, generated
"vision_tower.encoder.blocks.{bid}.wv", # kimi-vl, generated
"siglip2.vision_model.encoder.layers.{bid}.self_attn.v_proj",
),
MODEL_TENSOR.V_ENC_INPUT_NORM: (
@@ -1339,6 +1344,7 @@ class TensorNameMap:
"visual.blocks.{bid}.norm1", # qwen2vl
"vision_tower.encoder.blocks.{bid}.norm0", # kimi-vl (norm0/norm1)
"model.vision.transformer.layers.{bid}.input_layernorm", # cogvlm
"siglip2.vision_model.encoder.layers.{bid}.layer_norm1",
),
MODEL_TENSOR.V_ENC_ATTN_O: (
@@ -1354,6 +1360,7 @@ class TensorNameMap:
"visual.blocks.{bid}.attn.proj", # qwen2vl
"vision_tower.encoder.blocks.{bid}.wo", # kimi-vl
"model.vision.transformer.layers.{bid}.attention.dense", # cogvlm
"siglip2.vision_model.encoder.layers.{bid}.self_attn.out_proj", # youtuvl
),
MODEL_TENSOR.V_ENC_POST_ATTN_NORM: (
@@ -1368,6 +1375,7 @@ class TensorNameMap:
"visual.blocks.{bid}.norm2", # qwen2vl
"vision_tower.encoder.blocks.{bid}.norm1", # kimi-vl (norm0/norm1)
"model.vision.transformer.layers.{bid}.post_attention_layernorm", # cogvlm
"siglip2.vision_model.encoder.layers.{bid}.layer_norm2",
),
MODEL_TENSOR.V_ENC_FFN_UP: (
@@ -1383,6 +1391,7 @@ class TensorNameMap:
"visual.blocks.{bid}.mlp.linear_fc1", # qwen3vl
"vision_tower.encoder.blocks.{bid}.mlp.fc0", # kimi-vl (fc0/fc1)
"model.vision.transformer.layers.{bid}.mlp.fc1", # cogvlm
"siglip2.vision_model.encoder.layers.{bid}.mlp.fc1",
),
MODEL_TENSOR.V_ENC_FFN_GATE: (
@@ -1404,6 +1413,7 @@ class TensorNameMap:
"visual.blocks.{bid}.mlp.linear_fc2", # qwen3vl
"vision_tower.encoder.blocks.{bid}.mlp.fc1", # kimi-vl (fc0/fc1)
"model.vision.transformer.layers.{bid}.mlp.fc2", # cogvlm
"siglip2.vision_model.encoder.layers.{bid}.mlp.fc2",
),
MODEL_TENSOR.V_LAYER_SCALE_1: (
@@ -1430,6 +1440,7 @@ class TensorNameMap:
"visual.merger.ln_q", # qwen2vl
"vision_tower.encoder.final_layernorm", # kimi-vl
"visual.post_layernorm", # glm4v
"siglip2.vision_model.post_layernorm",
),
MODEL_TENSOR.V_MM_POST_NORM: (
@@ -1446,6 +1457,7 @@ class TensorNameMap:
"multi_modal_projector.pre_norm",
"pre_mm_projector_norm",
"model.vision.linear_proj.norm1", # cogvlm
"merger.ln_q",
),
MODEL_TENSOR.V_MM_SOFT_EMB_NORM: (