ver4a/llama.cpp
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity

convert : bailingmoe : set yarn metadata if present (#13312)

Browse source
This commit is contained in:
Sigbjørn Skjæret 2025-05-05 12:34:26 +02:00 • committed by GitHub
parent 66645a5285
commit ae803bfc3d
No known key found for this signature in database
GPG key ID: B5690EEEBB952194
1 changed files with 6 additions and 1 deletions
Download patch file Download diff file Expand all files Collapse all files

7
convert_hf_to_gguf.py
View file

@ -5680,7 +5680,12 @@ class BailingMoeModel(TextModel):
rope_dim = hparams.get("head_dim") or hparams["hidden_size"] // hparams["num_attention_heads"]
self.gguf_writer.add_rope_dimension_count(rope_dim)
self.gguf_writer.add_rope_scaling_type(gguf.RopeScalingType.NONE)
if (self.hparams.get("rope_scaling") or {}).get("type") == "yarn" and "factor" in self.hparams["rope_scaling"]:
self.gguf_writer.add_rope_scaling_type(gguf.RopeScalingType.YARN)
self.gguf_writer.add_rope_scaling_factor(self.hparams["rope_scaling"]["factor"])
self.gguf_writer.add_rope_scaling_orig_ctx_len(self.hparams["rope_scaling"]["original_max_position_embeddings"])
else:
self.gguf_writer.add_rope_scaling_type(gguf.RopeScalingType.NONE)
self.gguf_writer.add_leading_dense_block_count(hparams["first_k_dense_replace"])
self.gguf_writer.add_vocab_size(hparams["vocab_size"])
self.gguf_writer.add_expert_feed_forward_length(hparams["moe_intermediate_size"])