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

Adding debug for 8 bit

Browse Source
This commit is contained in:
ebolam
2022-11-29 19:29:06 -05:00
parent be310bd10b
commit aabc73b3f7
1 changed files with 1 additions and 0 deletions
Download Patch File Download Diff File Expand all files Collapse all files

1
aiserver.py
View File

@@ -2948,6 +2948,7 @@ def load_model(use_gpu=True, gpu_layers=None, disk_layers=None, initial_load=Fal
except Exception as e:
tokenizer = GPT2Tokenizer.from_pretrained("gpt2", revision=koboldai_vars.revision, cache_dir="cache")
try:
logger.info("Using 8 bit: {}".format(use_8_bit))
model = AutoModelForCausalLM.from_pretrained(koboldai_vars.model, revision=koboldai_vars.revision, cache_dir="cache", load_in_8bit=use_8_bit, device_map="auto", **lowmem)
except Exception as e:
if("out of memory" in traceback.format_exc().lower()):