mirror of
https://github.com/bentoml/OpenLLM.git
synced 2026-06-12 10:29:36 -04:00
cron(style): run formatter [generated] [skip ci] (#257)
This commit is contained in:
@@ -1,5 +1,7 @@
|
||||
from __future__ import annotations
|
||||
import sys, typing as t
|
||||
import sys
|
||||
import typing as t
|
||||
|
||||
from openllm.exceptions import MissingDependencyError
|
||||
from openllm.utils import LazyModule, is_torch_available, is_vllm_available
|
||||
from openllm_core.config.configuration_starcoder import DEFAULT_PROMPT_TEMPLATE as DEFAULT_PROMPT_TEMPLATE, START_STARCODER_COMMAND_DOCSTRING as START_STARCODER_COMMAND_DOCSTRING, StarCoderConfig as StarCoderConfig
|
||||
|
||||
@@ -1,5 +1,9 @@
|
||||
from __future__ import annotations
|
||||
import logging, typing as t, bentoml, openllm
|
||||
import logging
|
||||
import typing as t
|
||||
|
||||
import bentoml
|
||||
import openllm
|
||||
from openllm.utils import generate_labels
|
||||
from openllm_core.config.configuration_starcoder import EOD, FIM_MIDDLE, FIM_PAD, FIM_PREFIX, FIM_SUFFIX
|
||||
if t.TYPE_CHECKING: import transformers
|
||||
@@ -12,7 +16,8 @@ class StarCoder(openllm.LLM['transformers.GPTBigCodeForCausalLM', 'transformers.
|
||||
return {'device_map': 'auto' if torch.cuda.is_available() and torch.cuda.device_count() > 1 else None, 'torch_dtype': torch.float16 if torch.cuda.is_available() else torch.float32}, {}
|
||||
|
||||
def import_model(self, *args: t.Any, trust_remote_code: bool = False, **attrs: t.Any) -> bentoml.Model:
|
||||
import torch, transformers
|
||||
import torch
|
||||
import transformers
|
||||
torch_dtype, device_map = attrs.pop('torch_dtype', torch.float16), attrs.pop('device_map', 'auto')
|
||||
tokenizer = transformers.AutoTokenizer.from_pretrained(self.model_id, **self.llm_parameters[-1])
|
||||
tokenizer.add_special_tokens({'additional_special_tokens': [EOD, FIM_PREFIX, FIM_MIDDLE, FIM_SUFFIX, FIM_PAD], 'pad_token': EOD})
|
||||
|
||||
@@ -1,5 +1,8 @@
|
||||
from __future__ import annotations
|
||||
import logging, typing as t, openllm
|
||||
import logging
|
||||
import typing as t
|
||||
|
||||
import openllm
|
||||
if t.TYPE_CHECKING: import vllm, transformers
|
||||
class VLLMStarCoder(openllm.LLM['vllm.LLMEngine', 'transformers.GPT2TokenizerFast']):
|
||||
__openllm_internal__ = True
|
||||
|
||||
Reference in New Issue
Block a user