Skip to content

merge google/flan based adapters: T5Adapter, CodeT5pAdapter, FlanAdapter #2411

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
merrymercy merged 9 commits into from
Sep 18, 2023
Merged

merge google/flan based adapters: T5Adapter, CodeT5pAdapter, FlanAdapter #2411

Changes from 1 commit
Commits
Show all changes
9 commits
Select commit Hold shift + click to select a range
120f58a
merge google/flan based adapters: T5Adapter, CodeT5pAdapter, FlanAdapter
wangzhen263 Sep 13, 2023
ed5bbb6
more specific checking on model name
wangzhen263 Sep 13, 2023
d413f5c
Merge branch 'lm-sys:main' into main
wangzhen263 Sep 13, 2023
9ce4681
fix code style
wangzhen263 Sep 13, 2023
1b0d4c5
Merge branch 'main' of https://github.com/wangzhen263/FastChat into main
wangzhen263 Sep 13, 2023
5881a95
Revert "fix code style"
wangzhen263 Sep 13, 2023
a3e2533
fix code style
wangzhen263 Sep 13, 2023
3cf629b
Update fastchat/model/model_adapter.py
wangzhen263 Sep 18, 2023
a7a4bc0
reformat model_adapter.py for the new commit
wangzhen263 Sep 18, 2023
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
Prev Previous commit
Next Next commit
Revert "fix code style" 
This reverts commit 9ce4681.
  • Loading branch information
@wangzhen263
wangzhen263 committed Sep 13, 2023
commit 5881a9543758bf3bc10839d9cec78e1b86532496
53 changes: 38 additions & 15 deletions fastchat/model/model_adapter.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -30,7 +30,9 @@
from fastchat.modules.awq import AWQConfig, load_awq_quantized
from fastchat.conversation import Conversation, get_conv_template
from fastchat.model.compression import load_compress_model
from fastchat.model.llama_condense_monkey_patch import replace_llama_with_condense
from fastchat.model.llama_condense_monkey_patch import (
replace_llama_with_condense,
)
from fastchat.model.model_chatglm import generate_stream_chatglm
from fastchat.model.model_codet5p import generate_stream_codet5p
from fastchat.model.model_falcon import generate_stream_falcon
Expand Down Expand Up @@ -212,7 +214,7 @@ def load_model(

if "max_memory" in kwargs:
kwargs["max_memory"]["cpu"] = (
str(math.floor(psutil.virtual_memory().available / 2 ** 20)) + "Mib"
str(math.floor(psutil.virtual_memory().available / 2**20)) + "Mib"
)
kwargs["quantization_config"] = BitsAndBytesConfig(
load_in_8bit_fp32_cpu_offload=cpu_offloading
Expand Down Expand Up @@ -531,7 +533,9 @@ def load_model(self, model_path: str, from_pretrained_kwargs: dict):
model_path, use_fast=self.use_fast_tokenizer, revision=revision
)
model = AutoModelForCausalLM.from_pretrained(
model_path, low_cpu_mem_usage=True, **from_pretrained_kwargs,
model_path,
low_cpu_mem_usage=True,
**from_pretrained_kwargs,
)
self.raise_warning_for_old_weights(model)
return model, tokenizer
Expand Down Expand Up @@ -601,7 +605,9 @@ def load_model(self, model_path: str, from_pretrained_kwargs: dict):
model_path, use_fast=self.use_fast_tokenizer, revision=revision
)
model = AutoModelForCausalLM.from_pretrained(
model_path, low_cpu_mem_usage=True, **from_pretrained_kwargs,
model_path,
low_cpu_mem_usage=True,
**from_pretrained_kwargs,
)
return model, tokenizer

Expand All @@ -613,9 +619,7 @@ class GoogleFlanAdapter(BaseModelAdapter):
"""The model adapter for google/Flan based models, such as Salesforce/codet5p-6b, lmsys/fastchat-t5-3b-v1.0, flan-t5-*, flan-ul2"""

def match(self, model_path: str):
return any(
model_path in model_str for model_str in ["flan-", "fastchat-t5", "codet5p"]
)
return any(model_path in model_str for model_str in ["flan-", "fastchat-t5", "codet5p"])

def load_model(self, model_path: str, from_pretrained_kwargs: dict):
revision = from_pretrained_kwargs.get("revision", "main")
Expand Down Expand Up @@ -686,7 +690,9 @@ def load_model(self, model_path: str, from_pretrained_kwargs: dict):
revision = from_pretrained_kwargs.get("revision", "main")
tokenizer = AutoTokenizer.from_pretrained(model_path, revision=revision)
model = AutoModelForCausalLM.from_pretrained(
model_path, low_cpu_mem_usage=True, **from_pretrained_kwargs,
model_path,
low_cpu_mem_usage=True,
**from_pretrained_kwargs,
)
# 50277 means "### End"
tokenizer.eos_token_id = 50277
Expand Down Expand Up @@ -935,7 +941,9 @@ def load_model(self, model_path: str, from_pretrained_kwargs: dict):
revision = from_pretrained_kwargs.get("revision", "main")
tokenizer = AutoTokenizer.from_pretrained(model_path, revision=revision)
model = AutoModelForCausalLM.from_pretrained(
model_path, low_cpu_mem_usage=True, **from_pretrained_kwargs,
model_path,
low_cpu_mem_usage=True,
**from_pretrained_kwargs,
)
return model, tokenizer

Expand Down Expand Up @@ -1103,7 +1111,9 @@ def match(self, model_path: str):
def load_model(self, model_path: str, from_pretrained_kwargs: dict):
revision = from_pretrained_kwargs.get("revision", "main")
tokenizer = AutoTokenizer.from_pretrained(
model_path, trust_remote_code=True, revision=revision,
model_path,
trust_remote_code=True,
revision=revision,
)
model = AutoModelForCausalLM.from_pretrained(
model_path,
Expand Down Expand Up @@ -1267,7 +1277,9 @@ def load_model(self, model_path: str, from_pretrained_kwargs: dict):
model_path, use_fast=self.use_fast_tokenizer, revision=revision
)
model = AutoModelForCausalLM.from_pretrained(
model_path, low_cpu_mem_usage=True, **from_pretrained_kwargs,
model_path,
low_cpu_mem_usage=True,
**from_pretrained_kwargs,
).eval()
return model, tokenizer

Expand Down Expand Up @@ -1314,7 +1326,10 @@ def load_model(self, model_path: str, from_pretrained_kwargs: dict):
from transformers.generation import GenerationConfig

revision = from_pretrained_kwargs.get("revision", "main")
config = AutoConfig.from_pretrained(model_path, trust_remote_code=True,)
config = AutoConfig.from_pretrained(
model_path,
trust_remote_code=True,
)
# NOTE: if you use the old version of model file, please remove the comments below
# config.use_flash_attn = False
config.fp16 = True
Expand Down Expand Up @@ -1356,7 +1371,10 @@ def match(self, model_path: str):

def load_model(self, model_path: str, from_pretrained_kwargs: dict):
revision = from_pretrained_kwargs.get("revision", "main")
model = AutoModel.from_pretrained(model_path, **from_pretrained_kwargs,)
model = AutoModel.from_pretrained(
model_path,
**from_pretrained_kwargs,
)
tokenizer = AutoTokenizer.from_pretrained(
model_path, trust_remote_code=True, revision=revision
)
Expand All @@ -1382,7 +1400,10 @@ def match(self, model_path: str):

def load_model(self, model_path: str, from_pretrained_kwargs: dict):
revision = from_pretrained_kwargs.get("revision", "main")
model = AutoModel.from_pretrained(model_path, **from_pretrained_kwargs,)
model = AutoModel.from_pretrained(
model_path,
**from_pretrained_kwargs,
)
tokenizer = AutoTokenizer.from_pretrained(
model_path, trust_remote_code=True, revision=revision
)
Expand Down Expand Up @@ -1431,7 +1452,9 @@ def match(self, model_path: str):
def load_model(self, model_path: str, from_pretrained_kwargs: dict):
revision = from_pretrained_kwargs.get("revision", "main")
tokenizer = AutoTokenizer.from_pretrained(
model_path, trust_remote_code=True, revision=revision,
model_path,
trust_remote_code=True,
revision=revision,
)
model = AutoModelForCausalLM.from_pretrained(
model_path,
Expand Down