Spaces:
Runtime error
Runtime error
Wonderplex
commited on
sotopia-pi prompt template fix (#66)
Browse files
sotopia_pi_generate.py → sotopia_generate.py
RENAMED
@@ -3,6 +3,7 @@ import os
|
|
3 |
from typing import TypeVar
|
4 |
from functools import cache
|
5 |
import logging
|
|
|
6 |
|
7 |
import torch
|
8 |
from transformers import AutoModelForCausalLM, AutoTokenizer, pipeline
|
@@ -22,9 +23,10 @@ from langchain.prompts import (
|
|
22 |
PromptTemplate,
|
23 |
)
|
24 |
from langchain.schema import BaseOutputParser, OutputParserException
|
|
|
|
|
25 |
from message_classes import ActionType, AgentAction
|
26 |
from utils import format_docstring
|
27 |
-
|
28 |
from langchain_callback_handler import LoggingCallbackHandler
|
29 |
|
30 |
HF_TOKEN_KEY_FILE="./hf_token.key"
|
@@ -89,7 +91,7 @@ def prepare_model(model_name):
|
|
89 |
model = AutoModelForCausalLM.from_pretrained(
|
90 |
"mistralai/Mistral-7B-Instruct-v0.1",
|
91 |
cache_dir="./.cache",
|
92 |
-
device_map='cuda'
|
93 |
)
|
94 |
model = PeftModel.from_pretrained(model, model_name).to("cuda")
|
95 |
|
@@ -98,7 +100,7 @@ def prepare_model(model_name):
|
|
98 |
model = AutoModelForCausalLM.from_pretrained(
|
99 |
"mistralai/Mistral-7B-Instruct-v0.1",
|
100 |
cache_dir="./.cache",
|
101 |
-
device_map='cuda',
|
102 |
quantization_config=BitsAndBytesConfig(
|
103 |
load_in_4bit=True,
|
104 |
bnb_4bit_use_double_quant=True,
|
@@ -114,7 +116,7 @@ def prepare_model(model_name):
|
|
114 |
model = AutoModelForCausalLM.from_pretrained(
|
115 |
"mistralai/Mistral-7B-Instruct-v0.1",
|
116 |
cache_dir="./.cache",
|
117 |
-
device_map='cuda'
|
118 |
)
|
119 |
|
120 |
else:
|
@@ -131,7 +133,7 @@ def obtain_chain_hf(
|
|
131 |
max_tokens: int = 2700
|
132 |
) -> LLMChain:
|
133 |
human_message_prompt = HumanMessagePromptTemplate(
|
134 |
-
prompt=PromptTemplate(template=template, input_variables=input_variables)
|
135 |
)
|
136 |
chat_prompt_template = ChatPromptTemplate.from_messages([human_message_prompt])
|
137 |
model, tokenizer = prepare_model(model_name)
|
@@ -148,6 +150,7 @@ def obtain_chain_hf(
|
|
148 |
chain = LLMChain(llm=hf, prompt=chat_prompt_template)
|
149 |
return chain
|
150 |
|
|
|
151 |
def generate(
|
152 |
model_name: str,
|
153 |
template: str,
|
|
|
3 |
from typing import TypeVar
|
4 |
from functools import cache
|
5 |
import logging
|
6 |
+
import json
|
7 |
|
8 |
import torch
|
9 |
from transformers import AutoModelForCausalLM, AutoTokenizer, pipeline
|
|
|
23 |
PromptTemplate,
|
24 |
)
|
25 |
from langchain.schema import BaseOutputParser, OutputParserException
|
26 |
+
import spaces
|
27 |
+
|
28 |
from message_classes import ActionType, AgentAction
|
29 |
from utils import format_docstring
|
|
|
30 |
from langchain_callback_handler import LoggingCallbackHandler
|
31 |
|
32 |
HF_TOKEN_KEY_FILE="./hf_token.key"
|
|
|
91 |
model = AutoModelForCausalLM.from_pretrained(
|
92 |
"mistralai/Mistral-7B-Instruct-v0.1",
|
93 |
cache_dir="./.cache",
|
94 |
+
# device_map='cuda'
|
95 |
)
|
96 |
model = PeftModel.from_pretrained(model, model_name).to("cuda")
|
97 |
|
|
|
100 |
model = AutoModelForCausalLM.from_pretrained(
|
101 |
"mistralai/Mistral-7B-Instruct-v0.1",
|
102 |
cache_dir="./.cache",
|
103 |
+
# device_map='cuda',
|
104 |
quantization_config=BitsAndBytesConfig(
|
105 |
load_in_4bit=True,
|
106 |
bnb_4bit_use_double_quant=True,
|
|
|
116 |
model = AutoModelForCausalLM.from_pretrained(
|
117 |
"mistralai/Mistral-7B-Instruct-v0.1",
|
118 |
cache_dir="./.cache",
|
119 |
+
# device_map='cuda'
|
120 |
)
|
121 |
|
122 |
else:
|
|
|
133 |
max_tokens: int = 2700
|
134 |
) -> LLMChain:
|
135 |
human_message_prompt = HumanMessagePromptTemplate(
|
136 |
+
prompt=PromptTemplate(template="[INST] " + template + " [/INST]", input_variables=input_variables)
|
137 |
)
|
138 |
chat_prompt_template = ChatPromptTemplate.from_messages([human_message_prompt])
|
139 |
model, tokenizer = prepare_model(model_name)
|
|
|
150 |
chain = LLMChain(llm=hf, prompt=chat_prompt_template)
|
151 |
return chain
|
152 |
|
153 |
+
|
154 |
def generate(
|
155 |
model_name: str,
|
156 |
template: str,
|
sotopia_space/chat.py
CHANGED
@@ -6,7 +6,7 @@ from typing import Literal
|
|
6 |
import json
|
7 |
from collections import defaultdict
|
8 |
from utils import Environment, Agent, get_context_prompt, dialogue_history_prompt
|
9 |
-
from
|
10 |
from sotopia_space.constants import MODEL_OPTIONS
|
11 |
|
12 |
DEPLOYED = os.getenv("DEPLOYED", "true").lower() == "true"
|
|
|
6 |
import json
|
7 |
from collections import defaultdict
|
8 |
from utils import Environment, Agent, get_context_prompt, dialogue_history_prompt
|
9 |
+
from sotopia_generate import prepare_model, generate_action
|
10 |
from sotopia_space.constants import MODEL_OPTIONS
|
11 |
|
12 |
DEPLOYED = os.getenv("DEPLOYED", "true").lower() == "true"
|
sotopia_space/constants.py
CHANGED
@@ -5,7 +5,7 @@ MODEL_OPTIONS = [
|
|
5 |
"cmu-lti/sotopia-pi-mistral-7b-BC_SR",
|
6 |
"cmu-lti/sotopia-pi-mistral-7b-BC_SR_4bit",
|
7 |
"mistralai/Mistral-7B-Instruct-v0.1"
|
8 |
-
# "mistralai/Mixtral-8x7B-Instruct-v0.1",
|
9 |
# "togethercomputer/llama-2-7b-chat",
|
10 |
# "togethercomputer/llama-2-70b-chat",
|
11 |
# "togethercomputer/mpt-30b-chat",
|
|
|
5 |
"cmu-lti/sotopia-pi-mistral-7b-BC_SR",
|
6 |
"cmu-lti/sotopia-pi-mistral-7b-BC_SR_4bit",
|
7 |
"mistralai/Mistral-7B-Instruct-v0.1"
|
8 |
+
# "mistralai/Mixtral-8x7B-Instruct-v0.1", # TODO: Add these model
|
9 |
# "togethercomputer/llama-2-7b-chat",
|
10 |
# "togethercomputer/llama-2-70b-chat",
|
11 |
# "togethercomputer/mpt-30b-chat",
|