Align parameters for "max_token, repetition_penalty,presence_penalty,…

…frequency_penalty" (#608)

* align max_tokens

Signed-off-by: Xinyao Wang <[email protected]>

* aligin repetition_penalty

Signed-off-by: Xinyao Wang <[email protected]>

* fix bug

Signed-off-by: Xinyao Wang <[email protected]>

* align penalty parameters

Signed-off-by: Xinyao Wang <[email protected]>

* fix bug

Signed-off-by: Xinyao Wang <[email protected]>

* [pre-commit.ci] auto fixes from pre-commit.com hooks

for more information, see https://pre-commit.ci

* fix bug

Signed-off-by: Xinyao Wang <[email protected]>

* fix bug

Signed-off-by: Xinyao Wang <[email protected]>

* fix bug

Signed-off-by: Xinyao Wang <[email protected]>

* fix bug

Signed-off-by: Xinyao Wang <[email protected]>

* align max_tokens

Signed-off-by: Xinyao Wang <[email protected]>

* fix bug

Signed-off-by: Xinyao Wang <[email protected]>

* fix bug

Signed-off-by: Xinyao Wang <[email protected]>

* debug

Signed-off-by: Xinyao Wang <[email protected]>

* debug

Signed-off-by: Xinyao Wang <[email protected]>

* [pre-commit.ci] auto fixes from pre-commit.com hooks

for more information, see https://pre-commit.ci

* fix langchain version bug

Signed-off-by: Xinyao Wang <[email protected]>

* fix langchain version bug

Signed-off-by: Xinyao Wang <[email protected]>

* [pre-commit.ci] auto fixes from pre-commit.com hooks

for more information, see https://pre-commit.ci

---------

Signed-off-by: Xinyao Wang <[email protected]>
Co-authored-by: kevinintel <[email protected]>
Co-authored-by: pre-commit-ci[bot] <66853113+pre-commit-ci[bot]@users.noreply.github.com>
Co-authored-by: lvliang-intel <[email protected]>

comps/cores/mega/gateway.py

@@ -160,11 +160,13 @@ async def handle_request(self, request: Request):
         chat_request = ChatCompletionRequest.parse_obj(data)
         prompt = self._handle_message(chat_request.messages)
         parameters = LLMParams(
-            max_new_tokens=chat_request.max_tokens if chat_request.max_tokens else 1024,
+            max_tokens=chat_request.max_tokens if chat_request.max_tokens else 1024,
             top_k=chat_request.top_k if chat_request.top_k else 10,
             top_p=chat_request.top_p if chat_request.top_p else 0.95,
             temperature=chat_request.temperature if chat_request.temperature else 0.01,
-            repetition_penalty=chat_request.presence_penalty if chat_request.presence_penalty else 1.03,
+            frequency_penalty=chat_request.frequency_penalty if chat_request.frequency_penalty else 0.0,
+            presence_penalty=chat_request.presence_penalty if chat_request.presence_penalty else 0.0,
+            repetition_penalty=chat_request.repetition_penalty if chat_request.repetition_penalty else 1.03,
             streaming=stream_opt,
             chat_template=chat_request.chat_template if chat_request.chat_template else None,
         )
@@ -214,11 +216,13 @@ async def handle_request(self, request: Request):
         chat_request = ChatCompletionRequest.parse_obj(data)
         prompt = self._handle_message(chat_request.messages)
         parameters = LLMParams(
-            max_new_tokens=chat_request.max_tokens if chat_request.max_tokens else 1024,
+            max_tokens=chat_request.max_tokens if chat_request.max_tokens else 1024,
             top_k=chat_request.top_k if chat_request.top_k else 10,
             top_p=chat_request.top_p if chat_request.top_p else 0.95,
             temperature=chat_request.temperature if chat_request.temperature else 0.01,
-            repetition_penalty=chat_request.presence_penalty if chat_request.presence_penalty else 1.03,
+            frequency_penalty=chat_request.frequency_penalty if chat_request.frequency_penalty else 0.0,
+            presence_penalty=chat_request.presence_penalty if chat_request.presence_penalty else 0.0,
+            repetition_penalty=chat_request.repetition_penalty if chat_request.repetition_penalty else 1.03,
             streaming=stream_opt,
         )
         result_dict, runtime_graph = await self.megaservice.schedule(
@@ -350,11 +354,13 @@ async def handle_request(self, request: Request):
         chat_request = ChatCompletionRequest.parse_obj(data)
         prompt = self._handle_message(chat_request.messages)
         parameters = LLMParams(
-            max_new_tokens=chat_request.max_tokens if chat_request.max_tokens else 1024,
+            max_tokens=chat_request.max_tokens if chat_request.max_tokens else 1024,
             top_k=chat_request.top_k if chat_request.top_k else 10,
             top_p=chat_request.top_p if chat_request.top_p else 0.95,
             temperature=chat_request.temperature if chat_request.temperature else 0.01,
-            repetition_penalty=chat_request.presence_penalty if chat_request.presence_penalty else 1.03,
+            frequency_penalty=chat_request.frequency_penalty if chat_request.frequency_penalty else 0.0,
+            presence_penalty=chat_request.presence_penalty if chat_request.presence_penalty else 0.0,
+            repetition_penalty=chat_request.repetition_penalty if chat_request.repetition_penalty else 1.03,
             streaming=stream_opt,
         )
         result_dict, runtime_graph = await self.megaservice.schedule(
@@ -399,11 +405,13 @@ async def handle_request(self, request: Request):
         chat_request = AudioChatCompletionRequest.parse_obj(data)
         parameters = LLMParams(
             # relatively lower max_tokens for audio conversation
-            max_new_tokens=chat_request.max_tokens if chat_request.max_tokens else 128,
+            max_tokens=chat_request.max_tokens if chat_request.max_tokens else 128,
             top_k=chat_request.top_k if chat_request.top_k else 10,
             top_p=chat_request.top_p if chat_request.top_p else 0.95,
             temperature=chat_request.temperature if chat_request.temperature else 0.01,
-            repetition_penalty=chat_request.presence_penalty if chat_request.presence_penalty else 1.03,
+            frequency_penalty=chat_request.frequency_penalty if chat_request.frequency_penalty else 0.0,
+            presence_penalty=chat_request.presence_penalty if chat_request.presence_penalty else 0.0,
+            repetition_penalty=chat_request.repetition_penalty if chat_request.repetition_penalty else 1.03,
             streaming=False,  # TODO add streaming LLM output as input to TTS
         )
         result_dict, runtime_graph = await self.megaservice.schedule(
@@ -428,11 +436,13 @@ async def handle_request(self, request: Request):
         chat_request = ChatCompletionRequest.parse_obj(data)
         prompt = self._handle_message(chat_request.messages)
         parameters = LLMParams(
-            max_new_tokens=chat_request.max_tokens if chat_request.max_tokens else 1024,
+            max_tokens=chat_request.max_tokens if chat_request.max_tokens else 1024,
             top_k=chat_request.top_k if chat_request.top_k else 10,
             top_p=chat_request.top_p if chat_request.top_p else 0.95,
             temperature=chat_request.temperature if chat_request.temperature else 0.01,
-            repetition_penalty=chat_request.presence_penalty if chat_request.presence_penalty else 1.03,
+            frequency_penalty=chat_request.frequency_penalty if chat_request.frequency_penalty else 0.0,
+            presence_penalty=chat_request.presence_penalty if chat_request.presence_penalty else 0.0,
+            repetition_penalty=chat_request.repetition_penalty if chat_request.repetition_penalty else 1.03,
             streaming=stream_opt,
         )
         result_dict, runtime_graph = await self.megaservice.schedule(
@@ -472,11 +482,13 @@ async def handle_request(self, request: Request):
         chat_request = ChatCompletionRequest.parse_obj(data)
         prompt = self._handle_message(chat_request.messages)
         parameters = LLMParams(
-            max_new_tokens=chat_request.max_tokens if chat_request.max_tokens else 1024,
+            max_tokens=chat_request.max_tokens if chat_request.max_tokens else 1024,
             top_k=chat_request.top_k if chat_request.top_k else 10,
             top_p=chat_request.top_p if chat_request.top_p else 0.95,
             temperature=chat_request.temperature if chat_request.temperature else 0.01,
-            repetition_penalty=chat_request.presence_penalty if chat_request.presence_penalty else 1.03,
+            frequency_penalty=chat_request.frequency_penalty if chat_request.frequency_penalty else 0.0,
+            presence_penalty=chat_request.presence_penalty if chat_request.presence_penalty else 0.0,
+            repetition_penalty=chat_request.repetition_penalty if chat_request.repetition_penalty else 1.03,
             streaming=stream_opt,
         )
         result_dict, runtime_graph = await self.megaservice.schedule(
@@ -520,7 +532,9 @@ async def handle_request(self, request: Request):
             top_k=chat_request.top_k if chat_request.top_k else 10,
             top_p=chat_request.top_p if chat_request.top_p else 0.95,
             temperature=chat_request.temperature if chat_request.temperature else 0.01,
-            repetition_penalty=chat_request.presence_penalty if chat_request.presence_penalty else 1.03,
+            frequency_penalty=chat_request.frequency_penalty if chat_request.frequency_penalty else 0.0,
+            presence_penalty=chat_request.presence_penalty if chat_request.presence_penalty else 0.0,
+            repetition_penalty=chat_request.repetition_penalty if chat_request.repetition_penalty else 1.03,
             streaming=stream_opt,
         )
         result_dict, runtime_graph = await self.megaservice.schedule(
@@ -569,7 +583,9 @@ async def handle_request(self, request: Request):
             top_k=chat_request.top_k if chat_request.top_k else 10,
             top_p=chat_request.top_p if chat_request.top_p else 0.95,
             temperature=chat_request.temperature if chat_request.temperature else 0.01,
-            repetition_penalty=chat_request.presence_penalty if chat_request.presence_penalty else 1.03,
+            frequency_penalty=chat_request.frequency_penalty if chat_request.frequency_penalty else 0.0,
+            presence_penalty=chat_request.presence_penalty if chat_request.presence_penalty else 0.0,
+            repetition_penalty=chat_request.repetition_penalty if chat_request.repetition_penalty else 1.03,
             streaming=stream_opt,
         )
         result_dict, runtime_graph = await self.megaservice.schedule(
@@ -758,7 +774,9 @@ async def handle_request(self, request: Request):
             top_k=chat_request.top_k if chat_request.top_k else 10,
             top_p=chat_request.top_p if chat_request.top_p else 0.95,
             temperature=chat_request.temperature if chat_request.temperature else 0.01,
-            repetition_penalty=chat_request.presence_penalty if chat_request.presence_penalty else 1.03,
+            frequency_penalty=chat_request.frequency_penalty if chat_request.frequency_penalty else 0.0,
+            presence_penalty=chat_request.presence_penalty if chat_request.presence_penalty else 0.0,
+            repetition_penalty=chat_request.repetition_penalty if chat_request.repetition_penalty else 1.03,
             streaming=stream_opt,
             chat_template=chat_request.chat_template if chat_request.chat_template else None,
         )

comps/cores/proto/api_protocol.py

@@ -285,8 +285,9 @@ class AudioChatCompletionRequest(BaseModel):
     max_tokens: Optional[int] = 1024
     stop: Optional[Union[str, List[str]]] = None
     stream: Optional[bool] = False
-    presence_penalty: Optional[float] = 1.03
+    presence_penalty: Optional[float] = 0.0
     frequency_penalty: Optional[float] = 0.0
+    repetition_penalty: Optional[float] = 1.03
     user: Optional[str] = None
 
 
@@ -345,6 +346,7 @@ class CompletionRequest(BaseModel):
     echo: Optional[bool] = False
     presence_penalty: Optional[float] = 0.0
     frequency_penalty: Optional[float] = 0.0
+    repetition_penalty: Optional[float] = 1.03
     user: Optional[str] = None
     use_beam_search: Optional[bool] = False
     best_of: Optional[int] = None

comps/cores/proto/docarray.py

@@ -145,11 +145,14 @@ class RerankedDoc(BaseDoc):
 class LLMParamsDoc(BaseDoc):
     model: Optional[str] = None  # for openai and ollama
     query: str
+    max_tokens: int = 1024
     max_new_tokens: int = 1024
     top_k: int = 10
     top_p: float = 0.95
     typical_p: float = 0.95
     temperature: float = 0.01
+    frequency_penalty: float = 0.0
+    presence_penalty: float = 0.0
     repetition_penalty: float = 1.03
     streaming: bool = True
 
@@ -179,11 +182,14 @@ def chat_template_must_contain_variables(cls, v):
 
 
 class LLMParams(BaseDoc):
+    max_tokens: int = 1024
     max_new_tokens: int = 1024
     top_k: int = 10
     top_p: float = 0.95
     typical_p: float = 0.95
     temperature: float = 0.01
+    frequency_penalty: float = 0.0
+    presence_penalty: float = 0.0
     repetition_penalty: float = 1.03
     streaming: bool = True
 

comps/llms/faq-generation/tgi/langchain/llm.py

@@ -40,7 +40,7 @@ def llm_generate(input: LLMParamsDoc):
     llm_endpoint = os.getenv("TGI_LLM_ENDPOINT", "http://localhost:8080")
     llm = HuggingFaceEndpoint(
         endpoint_url=llm_endpoint,
-        max_new_tokens=input.max_new_tokens,
+        max_new_tokens=input.max_tokens,
         top_k=input.top_k,
         top_p=input.top_p,
         typical_p=input.typical_p,

comps/llms/faq-generation/tgi/langchain/requirements.txt

@@ -2,7 +2,10 @@ docarray[full]
 fastapi
 huggingface_hub
 langchain
+langchain-huggingface
+langchain-openai
 langchain_community
+langchainhub
 opentelemetry-api
 opentelemetry-exporter-otlp
 opentelemetry-sdk

comps/llms/summarization/tgi/langchain/llm.py

@@ -39,7 +39,7 @@ def llm_generate(input: LLMParamsDoc):
     llm_endpoint = os.getenv("TGI_LLM_ENDPOINT", "http://localhost:8080")
     llm = HuggingFaceEndpoint(
         endpoint_url=llm_endpoint,
-        max_new_tokens=input.max_new_tokens,
+        max_new_tokens=input.max_tokens,
         top_k=input.top_k,
         top_p=input.top_p,
         typical_p=input.typical_p,

comps/llms/text-generation/README.md

@@ -374,7 +374,7 @@ curl http://${your_ip}:8008/v1/chat/completions \
 
 ### 3.3 Consume LLM Service
 
-You can set the following model parameters according to your actual needs, such as `max_new_tokens`, `streaming`.
+You can set the following model parameters according to your actual needs, such as `max_tokens`, `streaming`.
 
 The `streaming` parameter determines the format of the data returned by the API. It will return text string with `streaming=false`, return text streaming flow with `streaming=true`.
 
@@ -385,7 +385,7 @@ curl http://${your_ip}:9000/v1/chat/completions \
   -H 'Content-Type: application/json' \
   -d '{
   "query":"What is Deep Learning?",
-  "max_new_tokens":17,
+  "max_tokens":17,
   "top_k":10,
   "top_p":0.95,
   "typical_p":0.95,
@@ -401,7 +401,7 @@ curl http://${your_ip}:9000/v1/chat/completions \
   -H 'Content-Type: application/json' \
   -d '{
   "query":"What is Deep Learning?",
-  "max_new_tokens":17,
+  "max_tokens":17,
   "top_k":10,
   "top_p":0.95,
   "typical_p":0.95,

comps/llms/text-generation/ollama/langchain/README.md

@@ -70,5 +70,5 @@ docker run --network host -e http_proxy=$http_proxy -e https_proxy=$https_proxy
 ## Consume the Ollama Microservice
 
 ```bash
-curl http://127.0.0.1:9000/v1/chat/completions  -X POST   -d '{"model": "llama3", "query":"What is Deep Learning?","max_new_tokens":32,"top_k":10,"top_p":0.95,"typical_p":0.95,"temperature":0.01,"repetition_penalty":1.03,"streaming":true}'   -H 'Content-Type: application/json'
+curl http://127.0.0.1:9000/v1/chat/completions  -X POST   -d '{"model": "llama3", "query":"What is Deep Learning?","max_tokens":32,"top_k":10,"top_p":0.95,"typical_p":0.95,"temperature":0.01,"repetition_penalty":1.03,"streaming":true}'   -H 'Content-Type: application/json'
 ```

comps/llms/text-generation/ollama/langchain/llm.py

@@ -25,7 +25,7 @@ def llm_generate(input: LLMParamsDoc):
     ollama = Ollama(
         base_url=ollama_endpoint,
         model=input.model if input.model else model_name,
-        num_predict=input.max_new_tokens,
+        num_predict=input.max_tokens,
         top_k=input.top_k,
         top_p=input.top_p,
         temperature=input.temperature,

comps/llms/text-generation/predictionguard/README.md

@@ -29,7 +29,7 @@ curl -X POST http://localhost:9000/v1/chat/completions \
     -d '{
         "model": "Hermes-2-Pro-Llama-3-8B",
         "query": "Tell me a joke.",
-        "max_new_tokens": 100,
+        "max_tokens": 100,
         "temperature": 0.7,
         "top_p": 0.9,
         "top_k": 50,
@@ -45,7 +45,7 @@ curl -N -X POST http://localhost:9000/v1/chat/completions \
     -d '{
         "model": "Hermes-2-Pro-Llama-3-8B",
         "query": "Tell me a joke.",
-        "max_new_tokens": 100,
+        "max_tokens": 100,
         "temperature": 0.7,
         "top_p": 0.9,
         "top_k": 50,

comps/llms/text-generation/predictionguard/llm_predictionguard.py

@@ -49,7 +49,7 @@ async def stream_generator():
             for res in client.chat.completions.create(
                 model=input.model,
                 messages=messages,
-                max_tokens=input.max_new_tokens,
+                max_tokens=input.max_tokens,
                 temperature=input.temperature,
                 top_p=input.top_p,
                 top_k=input.top_k,
@@ -69,7 +69,7 @@ async def stream_generator():
             response = client.chat.completions.create(
                 model=input.model,
                 messages=messages,
-                max_tokens=input.max_new_tokens,
+                max_tokens=input.max_tokens,
                 temperature=input.temperature,
                 top_p=input.top_p,
                 top_k=input.top_k,

comps/llms/text-generation/ray_serve/llm.py

@@ -47,7 +47,7 @@ def llm_generate(input: LLMParamsDoc):
         openai_api_base=llm_endpoint + "/v1",
         model_name=llm_model,
         openai_api_key=os.getenv("OPENAI_API_KEY", "not_needed"),
-        max_tokens=input.max_new_tokens,
+        max_tokens=input.max_tokens,
         temperature=input.temperature,
         streaming=input.streaming,
         request_timeout=600,

comps/llms/text-generation/tgi/README.md

@@ -88,36 +88,42 @@ curl http://${your_ip}:9000/v1/health_check\
 
 ### 3.2 Consume LLM Service
 
-You can set the following model parameters according to your actual needs, such as `max_new_tokens`, `streaming`.
+You can set the following model parameters according to your actual needs, such as `max_tokens`, `streaming`.
 
 The `streaming` parameter determines the format of the data returned by the API. It will return text string with `streaming=false`, return text streaming flow with `streaming=true`.
 
 ```bash
 # non-streaming mode
 curl http://${your_ip}:9000/v1/chat/completions \
   -X POST \
-  -d '{"query":"What is Deep Learning?","max_new_tokens":17,"top_k":10,"top_p":0.95,"typical_p":0.95,"temperature":0.01,"repetition_penalty":1.03,"streaming":false}' \
+  -d '{"query":"What is Deep Learning?","max_tokens":17,"top_k":10,"top_p":0.95,"typical_p":0.95,"temperature":0.01,"repetition_penalty":1.03,"streaming":false}' \
   -H 'Content-Type: application/json'
 
 # streaming mode
 curl http://${your_ip}:9000/v1/chat/completions \
   -X POST \
-  -d '{"query":"What is Deep Learning?","max_new_tokens":17,"top_k":10,"top_p":0.95,"typical_p":0.95,"temperature":0.01,"repetition_penalty":1.03,"streaming":true}' \
+  -d '{"query":"What is Deep Learning?","max_tokens":17,"top_k":10,"top_p":0.95,"typical_p":0.95,"temperature":0.01,"repetition_penalty":1.03,"streaming":true}' \
   -H 'Content-Type: application/json'
 
-# custom chat template
+# consume with SearchedDoc
 curl http://${your_ip}:9000/v1/chat/completions \
   -X POST \
-  -d '{"query":"What is Deep Learning?","max_new_tokens":17,"top_k":10,"top_p":0.95,"typical_p":0.95,"temperature":0.01,"repetition_penalty":1.03,"streaming":true, "chat_template":"### You are a helpful, respectful and honest assistant to help the user with questions.\n### Context: {context}\n### Question: {question}\n### Answer:"}' \
+  -d '{"initial_query":"What is Deep Learning?","retrieved_docs":[{"text":"Deep Learning is a ..."},{"text":"Deep Learning is b ..."}]}' \
   -H 'Content-Type: application/json'
+```
 
-# consume with SearchedDoc
+For parameters in above modes, please refer to [HuggingFace InferenceClient API](https://huggingface.co/docs/huggingface_hub/package_reference/inference_client#huggingface_hub.InferenceClient.text_generation) (except we rename 'max_new_tokens' to 'max_tokens')
+
+```bash
+# custom chat template
 curl http://${your_ip}:9000/v1/chat/completions \
   -X POST \
-  -d '{"initial_query":"What is Deep Learning?","retrieved_docs":[{"text":"Deep Learning is a ..."},{"text":"Deep Learning is b ..."}]}' \
+  -d '{"query":"What is Deep Learning?","max_tokens":17,"top_k":10,"top_p":0.95,"typical_p":0.95,"temperature":0.01,"presence_penalty":1.03", frequency_penalty":0.0, "streaming":true, "chat_template":"### You are a helpful, respectful and honest assistant to help the user with questions.\n### Context: {context}\n### Question: {question}\n### Answer:"}' \
   -H 'Content-Type: application/json'
 ```
 
+For parameters in Chat mode, please refer to [OpenAI API](https://platform.openai.com/docs/api-reference/chat/create)
+
 ### 4. Validated Model
 
 | Model                     | TGI |

comps/llms/text-generation/tgi/llm.py

@@ -69,7 +69,7 @@ async def llm_generate(input: Union[LLMParamsDoc, ChatCompletionRequest, Searche
         text_generation = await llm.text_generation(
             prompt=prompt,
             stream=new_input.streaming,
-            max_new_tokens=new_input.max_new_tokens,
+            max_new_tokens=new_input.max_tokens,
             repetition_penalty=new_input.repetition_penalty,
             temperature=new_input.temperature,
             top_k=new_input.top_k,
@@ -119,7 +119,7 @@ async def stream_generator():
         text_generation = await llm.text_generation(
             prompt=prompt,
             stream=input.streaming,
-            max_new_tokens=input.max_new_tokens,
+            max_new_tokens=input.max_tokens,
             repetition_penalty=input.repetition_penalty,
             temperature=input.temperature,
             top_k=input.top_k,