rixty/KoboldAI-Client
1
1
Fork 0
mirror of https://github.com/KoboldAI/KoboldAI-Client.git synced 2025-06-05 21:59:24 +02:00
Code Issues Projects Releases Wiki Activity

Remove except Exception so that errors actually show up

Browse Source
This commit is contained in:
0cc4m
2023-03-16 05:19:47 +00:00
parent b3b454bbe4
commit 5d17692c79
1 changed files with 7 additions and 12 deletions
Download Patch File Download Diff File Expand all files Collapse all files

19
aiserver.py
View File

@@ -3128,20 +3128,15 @@ def load_model(use_gpu=True, gpu_layers=None, disk_layers=None, initial_load=Fal
# tokenizer = GPT2Tokenizer.from_pretrained(koboldai_vars.custmodpth, revision=koboldai_vars.revision, cache_dir="cache") # tokenizer = GPT2Tokenizer.from_pretrained(koboldai_vars.custmodpth, revision=koboldai_vars.revision, cache_dir="cache")
# except Exception as e: # except Exception as e:
# tokenizer = GPT2Tokenizer.from_pretrained("gpt2", revision=koboldai_vars.revision, cache_dir="cache") # tokenizer = GPT2Tokenizer.from_pretrained("gpt2", revision=koboldai_vars.revision, cache_dir="cache")
try: # model = AutoModelForCausalLM.from_pretrained(koboldai_vars.custmodpth, revision=koboldai_vars.revision, cache_dir="cache", **lowmem)
# model = AutoModelForCausalLM.from_pretrained(koboldai_vars.custmodpth, revision=koboldai_vars.revision, cache_dir="cache", **lowmem)
if os.environ.get('LLAMA_4BIT') is not None:
model = load_quant(koboldai_vars.custmodpth, os.environ['LLAMA_4BIT'], 4)
else:
raise RuntimeError("It looks like your environment variable for LLAMA_4BIT is not set (the model path).\nPlease set this variable before proceeding.")
if model is None: if os.environ.get('LLAMA_4BIT'):
raise RuntimeError("Model returned 'None'. This is not expected to happen, but due to this, the model will not load.") model = load_quant(koboldai_vars.custmodpth, os.environ['LLAMA_4BIT'], 4)
else:
raise RuntimeError("It looks like your environment variable for LLAMA_4BIT is not set (the model path).\nPlease set this variable before proceeding.")
except Exception as e: if model is None:
if("out of memory" in traceback.format_exc().lower()): raise RuntimeError("Model returned 'None'. This is not expected to happen, but due to this, the model will not load.")
raise RuntimeError("One of your GPUs ran out of memory when KoboldAI tried to load your model.")
# model = GPTNeoForCausalLM.from_pretrained(koboldai_vars.custmodpth, revision=koboldai_vars.revision, cache_dir="cache", **lowmem)
elif(os.path.isdir("models/{}".format(koboldai_vars.model.replace('/', '_')))): elif(os.path.isdir("models/{}".format(koboldai_vars.model.replace('/', '_')))):
try: try:
tokenizer = AutoTokenizer.from_pretrained("models/{}".format(koboldai_vars.model.replace('/', '_')), revision=koboldai_vars.revision, cache_dir="cache", use_fast=False) tokenizer = AutoTokenizer.from_pretrained("models/{}".format(koboldai_vars.model.replace('/', '_')), revision=koboldai_vars.revision, cache_dir="cache", use_fast=False)