Husnain
commited on
Commit
•
a39e60f
1
Parent(s):
99c68b6
♻️ [Refactor] Move STOP_SEQUENCES_MAP and TOKEN_LIMIT_MAP to constants
Browse files- networks/message_streamer.py +22 -36
networks/message_streamer.py
CHANGED
@@ -1,49 +1,37 @@
|
|
1 |
import json
|
2 |
import re
|
3 |
import requests
|
|
|
4 |
from tiktoken import get_encoding as tiktoken_get_encoding
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
5 |
from messagers.message_outputer import OpenaiStreamOutputer
|
6 |
from utils.logger import logger
|
7 |
from utils.enver import enver
|
8 |
|
9 |
|
10 |
class MessageStreamer:
|
11 |
-
MODEL_MAP = {
|
12 |
-
"mixtral-8x7b": "mistralai/Mixtral-8x7B-Instruct-v0.1", # 72.62, fast [Recommended]
|
13 |
-
"mistral-7b": "mistralai/Mistral-7B-Instruct-v0.2", # 65.71, fast
|
14 |
-
"nous-mixtral-8x7b": "NousResearch/Nous-Hermes-2-Mixtral-8x7B-DPO",
|
15 |
-
"gemma-7b": "google/gemma-7b-it",
|
16 |
-
# "openchat-3.5": "openchat/openchat-3.5-1210", # 68.89, fast
|
17 |
-
# "zephyr-7b-beta": "HuggingFaceH4/zephyr-7b-beta", # ❌ Too Slow
|
18 |
-
# "llama-70b": "meta-llama/Llama-2-70b-chat-hf", # ❌ Require Pro User
|
19 |
-
# "codellama-34b": "codellama/CodeLlama-34b-Instruct-hf", # ❌ Low Score
|
20 |
-
# "falcon-180b": "tiiuae/falcon-180B-chat", # ❌ Require Pro User
|
21 |
-
"default": "mistralai/Mixtral-8x7B-Instruct-v0.1",
|
22 |
-
}
|
23 |
-
STOP_SEQUENCES_MAP = {
|
24 |
-
"mixtral-8x7b": "</s>",
|
25 |
-
"mistral-7b": "</s>",
|
26 |
-
"nous-mixtral-8x7b": "<|im_end|>",
|
27 |
-
"openchat-3.5": "<|end_of_turn|>",
|
28 |
-
"gemma-7b": "<eos>",
|
29 |
-
}
|
30 |
-
TOKEN_LIMIT_MAP = {
|
31 |
-
"mixtral-8x7b": 32768,
|
32 |
-
"mistral-7b": 32768,
|
33 |
-
"nous-mixtral-8x7b": 32768,
|
34 |
-
"openchat-3.5": 8192,
|
35 |
-
"gemma-7b": 8192,
|
36 |
-
}
|
37 |
-
TOKEN_RESERVED = 100
|
38 |
|
39 |
def __init__(self, model: str):
|
40 |
-
if model in
|
41 |
self.model = model
|
42 |
else:
|
43 |
self.model = "default"
|
44 |
-
self.model_fullname =
|
45 |
self.message_outputer = OpenaiStreamOutputer()
|
46 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
47 |
|
48 |
def parse_line(self, line):
|
49 |
line = line.decode("utf-8")
|
@@ -94,9 +82,7 @@ class MessageStreamer:
|
|
94 |
top_p = min(top_p, 0.99)
|
95 |
|
96 |
token_limit = int(
|
97 |
-
|
98 |
-
- self.TOKEN_RESERVED
|
99 |
-
- self.count_tokens(prompt) * 1.35
|
100 |
)
|
101 |
if token_limit <= 0:
|
102 |
raise ValueError("Prompt exceeded token limit!")
|
@@ -127,8 +113,8 @@ class MessageStreamer:
|
|
127 |
"stream": True,
|
128 |
}
|
129 |
|
130 |
-
if self.model in
|
131 |
-
self.stop_sequences =
|
132 |
# self.request_body["parameters"]["stop_sequences"] = [
|
133 |
# self.STOP_SEQUENCES[self.model]
|
134 |
# ]
|
@@ -178,7 +164,7 @@ class MessageStreamer:
|
|
178 |
logger.back(content, end="")
|
179 |
final_content += content
|
180 |
|
181 |
-
if self.model in
|
182 |
final_content = final_content.replace(self.stop_sequences, "")
|
183 |
|
184 |
final_content = final_content.strip()
|
|
|
1 |
import json
|
2 |
import re
|
3 |
import requests
|
4 |
+
|
5 |
from tiktoken import get_encoding as tiktoken_get_encoding
|
6 |
+
from transformers import AutoTokenizer
|
7 |
+
|
8 |
+
from constants.models import (
|
9 |
+
MODEL_MAP,
|
10 |
+
STOP_SEQUENCES_MAP,
|
11 |
+
TOKEN_LIMIT_MAP,
|
12 |
+
TOKEN_RESERVED,
|
13 |
+
)
|
14 |
from messagers.message_outputer import OpenaiStreamOutputer
|
15 |
from utils.logger import logger
|
16 |
from utils.enver import enver
|
17 |
|
18 |
|
19 |
class MessageStreamer:
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
20 |
|
21 |
def __init__(self, model: str):
|
22 |
+
if model in MODEL_MAP.keys():
|
23 |
self.model = model
|
24 |
else:
|
25 |
self.model = "default"
|
26 |
+
self.model_fullname = MODEL_MAP[self.model]
|
27 |
self.message_outputer = OpenaiStreamOutputer()
|
28 |
+
|
29 |
+
if self.model == "gemma-7b":
|
30 |
+
# this is not wrong, as repo `google/gemma-7b-it` is gated and must authenticate to access it
|
31 |
+
# so I use mistral-7b as a fallback
|
32 |
+
self.tokenizer = AutoTokenizer.from_pretrained(MODEL_MAP["mistral-7b"])
|
33 |
+
else:
|
34 |
+
self.tokenizer = AutoTokenizer.from_pretrained(self.model_fullname)
|
35 |
|
36 |
def parse_line(self, line):
|
37 |
line = line.decode("utf-8")
|
|
|
82 |
top_p = min(top_p, 0.99)
|
83 |
|
84 |
token_limit = int(
|
85 |
+
TOKEN_LIMIT_MAP[self.model] - TOKEN_RESERVED - self.count_tokens(prompt)
|
|
|
|
|
86 |
)
|
87 |
if token_limit <= 0:
|
88 |
raise ValueError("Prompt exceeded token limit!")
|
|
|
113 |
"stream": True,
|
114 |
}
|
115 |
|
116 |
+
if self.model in STOP_SEQUENCES_MAP.keys():
|
117 |
+
self.stop_sequences = STOP_SEQUENCES_MAP[self.model]
|
118 |
# self.request_body["parameters"]["stop_sequences"] = [
|
119 |
# self.STOP_SEQUENCES[self.model]
|
120 |
# ]
|
|
|
164 |
logger.back(content, end="")
|
165 |
final_content += content
|
166 |
|
167 |
+
if self.model in STOP_SEQUENCES_MAP.keys():
|
168 |
final_content = final_content.replace(self.stop_sequences, "")
|
169 |
|
170 |
final_content = final_content.strip()
|