LLama tokenizer workaround

This commit is contained in:
Henk
2023-03-06 19:32:28 +01:00
parent 1c891181e9
commit b7f30a41c5
2 changed files with 4 additions and 4 deletions

View File

@@ -3180,9 +3180,10 @@ def load_model(use_gpu=True, gpu_layers=None, disk_layers=None, initial_load=Fal
shutil.move(os.path.realpath(huggingface_hub.hf_hub_download(koboldai_vars.model, filename, revision=koboldai_vars.revision, cache_dir="cache", local_files_only=True, legacy_cache_layout=legacy)), os.path.join("models/{}".format(koboldai_vars.model.replace('/', '_')), filename))
shutil.rmtree("cache/")
if(koboldai_vars.badwordsids is koboldai_settings.badwordsids_default and koboldai_vars.model_type not in ("gpt2", "gpt_neo", "gptj")):
if(koboldai_vars.badwordsids is koboldai_settings.badwordsids_default and koboldai_vars.model_type not in ("gpt2", "gpt_neo", "gptj", "llama")):
koboldai_vars.badwordsids = [[v] for k, v in tokenizer.get_vocab().items() if any(c in str(k) for c in "<>[]") if koboldai_vars.newlinemode != "s" or str(k) != "</s>"]
if(koboldai_vars.badwordsids is koboldai_settings.badwordsids_default and koboldai_vars.model_type == "llama"):
koboldai_vars.badwordsids = koboldai_settings.badwordsids_llama
patch_causallm(model)
if(koboldai_vars.hascuda):