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

Update convert-llama-h5-to-gguf.py

Browse source
This commit is contained in:
klosax 2023-07-30 15:01:47 +02:00 • committed by GitHub
parent b19c11750b
commit 4ed98bf1ab
No known key found for this signature in database
GPG key ID: 4AEE18F83AFDEB23
1 changed files with 1 additions and 0 deletions
Download patch file Download diff file Expand all files Collapse all files

1
convert-llama-h5-to-gguf.py
View file

@ -87,6 +87,7 @@ gguf_writer.write_feed_forward_length(llm_arch, hparams["intermediate_size"])
gguf_writer.write_rope_dimension_count(llm_arch, hparams["hidden_size"] // hparams["num_attention_heads"]) gguf_writer.write_rope_dimension_count(llm_arch, hparams["hidden_size"] // hparams["num_attention_heads"])
gguf_writer.write_head_count(llm_arch, hparams["num_attention_heads"]) gguf_writer.write_head_count(llm_arch, hparams["num_attention_heads"])
gguf_writer.write_float32(llm_arch + ".attention.layer_norm_rms_epsilon", hparams["rms_norm_eps"]) gguf_writer.write_float32(llm_arch + ".attention.layer_norm_rms_epsilon", hparams["rms_norm_eps"])
gguf_writer.write_layer_norm_rms_eps(llm_arch, hparams["rms_norm_eps"])
# TOKENIZATION # TOKENIZATION