xu-song commited on
Commit
be9b2e2
·
1 Parent(s): 1aaa002

add deepseek

Browse files
Files changed (5) hide show
  1. chat_template_app.py +11 -0
  2. client.py +14 -0
  3. consistency_app.py +9 -0
  4. setup.md +5 -0
  5. utils/voting_util.py +15 -0
chat_template_app.py ADDED
@@ -0,0 +1,11 @@
 
 
 
 
 
 
 
 
 
 
 
 
1
+ """
2
+
3
+ ## chat template
4
+
5
+ - special_tokens
6
+ - default_system
7
+ - tools
8
+ - tool_call
9
+
10
+
11
+ """
client.py ADDED
@@ -0,0 +1,14 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ from gradio_client import Client
2
+
3
+
4
+ def self_chat_demo(system_message, num_turn=4):
5
+ client = Client("xu-song/tokenizer-arena")
6
+ result = client.predict(
7
+ text="Hello!!",
8
+ tokenizer_name="01-ai/Yi-1.5-34B",
9
+ api_name="/tokenize"
10
+ )
11
+ print(result)
12
+
13
+ if __name__ == "__main__":
14
+ self_chat_demo(system_message="你是一个小说家,擅长写武侠小说")
consistency_app.py ADDED
@@ -0,0 +1,9 @@
 
 
 
 
 
 
 
 
 
 
1
+ """
2
+ On the consistency of LLM tokenizer.
3
+
4
+
5
+
6
+
7
+ """
8
+
9
+
setup.md ADDED
@@ -0,0 +1,5 @@
 
 
 
 
 
 
1
+
2
+
3
+ ```sh
4
+ python compression_util.py
5
+ ```
utils/voting_util.py ADDED
@@ -0,0 +1,15 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ """
2
+ https://huggingface.co/spaces/open-llm-leaderboard/open_llm_leaderboard/blob/main/src/voting/vote_system.py
3
+
4
+
5
+ ## 原理
6
+
7
+ https://huggingface.co/docs/huggingface_hub/guides/upload
8
+
9
+ ## TODO
10
+ 投票需要增加哪些 tokenizer。
11
+ """
12
+
13
+
14
+ class VoteManager:
15
+ pass