root/llama.cpp
1
0
Fork 0
mirror of https://github.com/ggerganov/llama.cpp.git synced 2024-12-27 03:44:35 +00:00
Code Issues Actions 4 Packages Projects Releases Wiki Activity

convert-gptneox-h5-to-gguf.py : n_layer --> n_block

Browse Source
This commit is contained in:
klosax 2023-08-13 00:02:04 +02:00 committed by GitHub
parent d2ce9cfe8d
commit 489616e126
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23
1 changed files with 2 additions and 2 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

4
convert-gptneox-h5-to-gguf.py
View File

@ -85,7 +85,7 @@ gguf_writer.add_name(last_dir)
gguf_writer.add_architecture(llm_arch)
gguf_writer.add_context_length(llm_arch, hparams["max_position_embeddings"])
gguf_writer.add_embedding_length(llm_arch, hparams["hidden_size"])
gguf_writer.add_layer_count(llm_arch, block_count)
gguf_writer.add_block_count(llm_arch, block_count)
gguf_writer.add_feed_forward_length(llm_arch, hparams["intermediate_size"])
gguf_writer.add_rope_dimension_count(llm_arch, int( hparams["rotary_pct"]*(hparams["hidden_size"]//hparams["num_attention_heads"])) )
gguf_writer.add_head_count(llm_arch, hparams["num_attention_heads"])
@ -116,7 +116,7 @@ if Path(dir_model + "/tokenizer.json").is_file():
vocab_size = len( tokenizer_json["model"]["vocab"] )
# from ggllm.cpp falcon_convert.py
# ref: https://github.com/cmp-nct/ggllm.cpp/blob/master/falcon_convert.py
tokenizer = AutoTokenizer.from_pretrained(dir_model)
reverse_vocab = {id: encoded_tok for encoded_tok, id in tokenizer.vocab.items()}