search

ggerganov / llama.cpp

LLM inference in C/C++
MIT License
68.06k stars 9.76k forks source link

Freshly converted PLaMo fails assertion: vocab.id_to_token.size() == vocab.token_to_id.size() #5669

Closed cebtenzzre closed 1 month ago

cebtenzzre commented 9 months ago

Steps to Reproduce

  1. Download pfnet/plamo-13b-instruct
  2. Convert with convert-hf-to-gguf.py
  3. Attempt to run inference with main

Fails with:

GGML_ASSERT: /home/jared/src/forks/llama.cpp-2/llama.cpp:3395: vocab.id_to_token.size() == vocab.token_to_id.size()
github-actions[bot] commented 7 months ago

This issue was closed because it has been inactive for 14 days since being marked as stale.

cebtenzzre commented 7 months ago

I'm not aware of any recent development of the PLaMo code that would make this issue obsolete.

github-actions[bot] commented 5 months ago

This issue was closed because it has been inactive for 14 days since being marked as stale.

github-actions[bot] commented 4 months ago

This issue was closed because it has been inactive for 14 days since being marked as stale.

espoirMur commented 4 months ago

Shouldn't be closed. I have got the same issues with multilingual model croissantllm/CroissantLLMChat-v0.1 still investigating

stanus74 commented 2 months ago

I have the same problem. converted according to these instructions: https://github.com/ggerganov/llama.cpp/discussions/7927

ERROR [apps.rag.main] Ollama: llama runner process has terminated: GGML_ASSERT(vocab.id_to_token.size() == vocab.token_to_id.size()) failed

this embedding model:

https://huggingface.co/aari1995/German_Semantic_V3b

with this model the conversion worked without errors :

https://huggingface.co/danielheinz/e5-base-sts-en-de

github-actions[bot] commented 1 month ago

This issue was closed because it has been inactive for 14 days since being marked as stale.