NousResearch/atropos
1
0
Fork 0
mirror of https://github.com/NousResearch/atropos.git synced 2026-04-19 12:57:58 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions

add tokenizer name config to set the vllm/sglang tokenizer to something different if needed

Browse source
This commit is contained in:
Dakota 2026-02-09 15:26:29 -06:00
parent 13f282aabc
commit 7d6aeb9bbf
4 changed files with 17 additions and 2 deletions
Download patch file Download diff file Expand all files Collapse all files

4
atroposlib/envs/server_handling/server_baseline.py
View file

@ -250,6 +250,10 @@ class ServerBaseline(BaseModel):
server_type: Literal["openai", "trl", "sglang", "vllm"] = Field(
default="openai", description="Type of server to use"
)
tokenizer_name: str = Field(
default="none",
description="The tokenizer name to use. If none, will use the model_name as the tokenizer.",
)
class APIServerConfig(ServerBaseline):

1
atroposlib/envs/server_handling/server_manager.py
View file

@ -121,6 +121,7 @@ class ServerManager:
model_name=configs.model_name,
rolling_buffer_length=configs.rolling_buffer_length,
api_key="x",
tokenizer_name=configs.tokenizer_name,
)
)
self.servers = [

7
atroposlib/envs/server_handling/sglang_server.py
View file

@ -31,7 +31,12 @@ class SGLangServer(APIServer):
base_url=config.base_url,
timeout=config.timeout,
)
self.tokenizer = AutoTokenizer.from_pretrained(config.model_name)
tokenizer_name = (
config.model_name
if config.tokenizer_name == "none"
else config.tokenizer_name
)
self.tokenizer = AutoTokenizer.from_pretrained(tokenizer_name)
super().__init__(config, reasoning_config=reasoning_config)
async def check_server_status_task(self, chat_completion: bool = True):

7
atroposlib/envs/server_handling/vllm_server.py
View file

@ -34,7 +34,12 @@ class VLLMServer(APIServer):
base_url=config.base_url,
timeout=config.timeout,
)
self.tokenizer = AutoTokenizer.from_pretrained(config.model_name)
tokenizer_name = (
config.model_name
if config.tokenizer_name == "none"
else config.tokenizer_name
)
self.tokenizer = AutoTokenizer.from_pretrained(tokenizer_name)
super().__init__(config, reasoning_config=reasoning_config)
async def check_server_status_task(self, chat_completion: bool = True):