mirror of
				https://github.com/ggml-org/llama.cpp.git
				synced 2025-10-31 08:51:55 +00:00 
			
		
		
		
	Update convert-llama-h5-to-gguf.py
This commit is contained in:
		| @@ -91,11 +91,13 @@ gguf_writer.write_float32(llm_arch + ".attention.layer_norm_rms_epsilon", hparam | |||||||
|  |  | ||||||
| # TOKENIZATION | # TOKENIZATION | ||||||
|  |  | ||||||
|  | print("write gguf tokenizer") | ||||||
|  |  | ||||||
| tokens: List[str] = [] | tokens: List[str] = [] | ||||||
| scores: List[float] = [] | scores: List[float] = [] | ||||||
|  |  | ||||||
| if Path(dir_model + "/tokenizer.model").is_file(): | if Path(dir_model + "/tokenizer.model").is_file(): | ||||||
|     # vocab type SPIECE |     # vocab type sentencepiece | ||||||
|     print("Adding sentencepiece tokenizer vocab.") |     print("Adding sentencepiece tokenizer vocab.") | ||||||
|     tokenizer = SentencePieceProcessor(dir_model + "/tokenizer.model") |     tokenizer = SentencePieceProcessor(dir_model + "/tokenizer.model") | ||||||
|  |  | ||||||
| @@ -123,15 +125,12 @@ if Path(dir_model + "/tokenizer.model").is_file(): | |||||||
|         tokens.append(text) |         tokens.append(text) | ||||||
|         scores.append(score) |         scores.append(score) | ||||||
|  |  | ||||||
| print("write gguf tokens") |  | ||||||
|  |  | ||||||
| gguf_writer.write_tokenizer_model("llama") | gguf_writer.write_tokenizer_model("llama") | ||||||
| gguf_writer.write_token_list(tokens) | gguf_writer.write_token_list(tokens) | ||||||
| gguf_writer.write_token_scores(scores) | gguf_writer.write_token_scores(scores) | ||||||
|  |  | ||||||
| # TENSORS | # TENSORS | ||||||
|  |  | ||||||
|  |  | ||||||
| # tensor info | # tensor info | ||||||
| print("write gguf tensor info") | print("write gguf tensor info") | ||||||
|  |  | ||||||
|   | |||||||
		Reference in New Issue
	
	Block a user
	 klosax
					klosax