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

Also fallback to generic GPT2 tokenizer in Colab TPU instances

This commit is contained in:
Gnome Ann 2022-03-12 23:19:35 -05:00
parent d29a629320
commit 48d07adb54
1 changed files with 18 additions and 9 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

27
tpu_mtj_backend.py
View File

@ -1035,29 +1035,38 @@ def load_model(path: str, driver_version="tpu_driver0.1_dev20210607", hf_checkpo
if(os.path.isdir(vars.custmodpth)): if(os.path.isdir(vars.custmodpth)):
try: try:
tokenizer = AutoTokenizer.from_pretrained(vars.custmodpth, cache_dir="cache") tokenizer = AutoTokenizer.from_pretrained(vars.custmodpth, cache_dir="cache")
except ValueError as e: except Exception as e:
tokenizer = GPT2TokenizerFast.from_pretrained(vars.custmodpth, cache_dir="cache") try:
tokenizer = GPT2TokenizerFast.from_pretrained(vars.custmodpth, cache_dir="cache")
except Exception as e:
tokenizer = GPT2TokenizerFast.from_pretrained("gpt2", cache_dir="cache")
try: try:
model = AutoModelForCausalLM.from_pretrained(vars.custmodpth, cache_dir="cache") model = AutoModelForCausalLM.from_pretrained(vars.custmodpth, cache_dir="cache")
except ValueError as e: except Exception as e:
model = GPTNeoForCausalLM.from_pretrained(vars.custmodpth, cache_dir="cache") model = GPTNeoForCausalLM.from_pretrained(vars.custmodpth, cache_dir="cache")
elif(os.path.isdir("models/{}".format(vars.model.replace('/', '_')))): elif(os.path.isdir("models/{}".format(vars.model.replace('/', '_')))):
try: try:
tokenizer = AutoTokenizer.from_pretrained("models/{}".format(vars.model.replace('/', '_')), cache_dir="cache") tokenizer = AutoTokenizer.from_pretrained("models/{}".format(vars.model.replace('/', '_')), cache_dir="cache")
except ValueError as e: except Exception as e:
tokenizer = GPT2TokenizerFast.from_pretrained("models/{}".format(vars.model.replace('/', '_')), cache_dir="cache") try:
tokenizer = GPT2TokenizerFast.from_pretrained("models/{}".format(vars.model.replace('/', '_')), cache_dir="cache")
except Exception as e:
tokenizer = GPT2TokenizerFast.from_pretrained("gpt2", cache_dir="cache")
try: try:
model = AutoModelForCausalLM.from_pretrained("models/{}".format(vars.model.replace('/', '_')), cache_dir="cache") model = AutoModelForCausalLM.from_pretrained("models/{}".format(vars.model.replace('/', '_')), cache_dir="cache")
except ValueError as e: except Exception as e:
model = GPTNeoForCausalLM.from_pretrained("models/{}".format(vars.model.replace('/', '_')), cache_dir="cache") model = GPTNeoForCausalLM.from_pretrained("models/{}".format(vars.model.replace('/', '_')), cache_dir="cache")
else: else:
try: try:
tokenizer = AutoTokenizer.from_pretrained(vars.model, cache_dir="cache") tokenizer = AutoTokenizer.from_pretrained(vars.model, cache_dir="cache")
except ValueError as e: except Exception as e:
tokenizer = GPT2TokenizerFast.from_pretrained(vars.model, cache_dir="cache") try:
tokenizer = GPT2TokenizerFast.from_pretrained(vars.model, cache_dir="cache")
except Exception as e:
tokenizer = GPT2TokenizerFast.from_pretrained("gpt2", cache_dir="cache")
try: try:
model = AutoModelForCausalLM.from_pretrained(vars.model, cache_dir="cache") model = AutoModelForCausalLM.from_pretrained(vars.model, cache_dir="cache")
except ValueError as e: except Exception as e:
model = GPTNeoForCausalLM.from_pretrained(vars.model, cache_dir="cache") model = GPTNeoForCausalLM.from_pretrained(vars.model, cache_dir="cache")
network.state = network.move_xmap(network.state, np.zeros(cores_per_replica)) network.state = network.move_xmap(network.state, np.zeros(cores_per_replica))