rixty
/
KoboldAI-Client
mirror of https://github.com/KoboldAI/KoboldAI-Client.git
1
1
Fork 0
Code Issues Projects Releases Wiki Activity

Merge pull request #223 from ebolam/united 

Fix for GPT models downloading even when present in model folder
This commit is contained in:
henk717 2022-09-28 19:15:33 +02:00 committed by GitHub
parent 0f7ecb3257 e7973e13ac
commit 3a094a049b
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23
1 changed files with 12 additions and 2 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

14
aiserver.py
View File

@ -2442,15 +2442,25 @@ def load_model(use_gpu=True, gpu_layers=None, disk_layers=None, initial_load=Fal
config_path = os.path.join("models/", vars.custmodpth)
config_path = os.path.join(config_path, "config.json").replace("\\", "//")
model_config = open(config_path, "r")
js = json.load(model_config)
#js = json.load(model_config)
with(maybe_use_float16()):
try:
model = GPT2LMHeadModel.from_pretrained(vars.custmodpth, revision=vars.revision, cache_dir="cache")
if os.path.exists(vars.custmodpth):
model = GPT2LMHeadModel.from_pretrained(vars.custmodpth, revision=vars.revision, cache_dir="cache")
tokenizer = GPT2Tokenizer.from_pretrained(vars.custmodpth, revision=vars.revision, cache_dir="cache")
elif os.path.exists(os.path.join("models/", vars.custmodpth)):
model = GPT2LMHeadModel.from_pretrained(os.path.join("models/", vars.custmodpth), revision=vars.revision, cache_dir="cache")
tokenizer = GPT2Tokenizer.from_pretrained(os.path.join("models/", vars.custmodpth), revision=vars.revision, cache_dir="cache")
else:
model = GPT2LMHeadModel.from_pretrained(vars.custmodpth, revision=vars.revision, cache_dir="cache")
tokenizer = GPT2Tokenizer.from_pretrained(vars.custmodpth, revision=vars.revision, cache_dir="cache")
except Exception as e:
if("out of memory" in traceback.format_exc().lower()):
raise RuntimeError("One of your GPUs ran out of memory when KoboldAI tried to load your model.")
raise e
tokenizer = GPT2Tokenizer.from_pretrained(vars.custmodpth, revision=vars.revision, cache_dir="cache")
model.save_pretrained("models/{}".format(vars.model.replace('/', '_')), max_shard_size="500MiB")
tokenizer.save_pretrained("models/{}".format(vars.model.replace('/', '_')))
vars.modeldim = get_hidden_size_from_model(model)
# Is CUDA available? If so, use GPU, otherwise fall back to CPU
if(vars.hascuda and vars.usegpu):