CS348Project/llama.cpp
5
0
Fork 0
mirror of https://github.com/ggml-org/llama.cpp.git synced 2025-10-27 08:21:30 +00:00
Code Issues Packages Projects Releases Wiki Activity

split q_proj/gate

Browse Source
This commit is contained in:
Xuan Son Nguyen
2025-09-20 12:00:14 +07:00
parent 178230ee21
commit 46110e0630
6 changed files with 31 additions and 21 deletions
Download Patch File Download Diff File Expand all files Collapse all files

6
convert_hf_to_gguf.py
View File

@@ -3767,8 +3767,12 @@ class Qwen3NextModel(Qwen3MoeModel):
name = name.rpartition(".dt_bias")[0] + ".dt_proj.bias"
elif "conv1d" in name:
data_torch = data_torch.squeeze()
elif "q_proj.weight" in name:
q_proj, gate = data_torch.chunk(2, dim=0)
yield (self.format_tensor_name(gguf.MODEL_TENSOR.ATTN_GATE, bid), gate)
data_torch = q_proj
return Qwen2MoeModel.modify_tensors(self, data_torch, name, bid)
yield from Qwen2MoeModel.modify_tensors(self, data_torch, name, bid)
@ModelBase.register("GPT2LMHeadModel")