CS348Project/llama.cpp
5
0
Fork 0
mirror of https://github.com/ggml-org/llama.cpp.git synced 2025-10-30 08:42:00 +00:00
Code Issues Packages Projects Releases 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
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"] 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_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_leading_dense_block_count(hparams["first_k_dense_replace"])
self.gguf_writer.add_vocab_size(hparams["vocab_size"]) self.gguf_writer.add_vocab_size(hparams["vocab_size"])
self.gguf_writer.add_expert_feed_forward_length(hparams["moe_intermediate_size"]) self.gguf_writer.add_expert_feed_forward_length(hparams["moe_intermediate_size"])