Merge pull request #16 from justinh-rahb/mlx-lm

Fix MLX Pipeline

pipelines/examples/mlx_pipeline.py

@@ -1,118 +1,99 @@
 """
 title: MLX Pipeline
 author: justinh-rahb
-date: 2024-05-22
+date: 2024-05-27
-version: 1.0
+version: 1.1
 license: MIT
-description: A pipeline for running the mlx-lm server with a specified model.
+description: A pipeline for generating text using Apple MLX Framework.
-dependencies: requests, mlx-lm
+dependencies: requests, mlx-lm, huggingface-hub
-environment_variables: MLX_MODEL
+environment_variables: MLX_HOST, MLX_PORT, MLX_MODEL, MLX_STOP, MLX_SUBPROCESS, HUGGINGFACE_TOKEN
 """
 
 from typing import List, Union, Generator, Iterator
-import requests
-import subprocess
-import os
-import socket
 from schemas import OpenAIChatMessage
-
+import requests
+import os
+import subprocess
+import logging
+from huggingface_hub import login
 
 class Pipeline:
     def __init__(self):
-        # Optionally, you can set the id and name of the pipeline.
         self.id = "mlx_pipeline"
         self.name = "MLX Pipeline"
-        self.process = None
+        self.host = os.getenv("MLX_HOST", "localhost")
-        self.model = os.getenv(
+        self.port = os.getenv("MLX_PORT", "8080")
-            "MLX_MODEL", "mistralai/Mistral-7B-Instruct-v0.2"
+        self.model = os.getenv("MLX_MODEL", "mistralai/Mistral-7B-Instruct-v0.2")
-        )  # Default model if not set in environment variable
+        self.stop_sequence = os.getenv("MLX_STOP", "[INST]").split(",")  # Default stop sequence is [INST]
-        self.port = self.find_free_port()
+        self.subprocess = os.getenv("MLX_SUBPROCESS", "true").lower() == "true"
-        self.stop_sequences = os.getenv(
+        self.huggingface_token = os.getenv("HUGGINGFACE_TOKEN", None)
-            "MLX_STOP", "[INST]"
-        )  # Stop sequences from environment variable
 
-    @staticmethod
+        if self.huggingface_token:
-    def find_free_port():
+            login(self.huggingface_token)
-        with socket.socket(socket.AF_INET, socket.SOCK_STREAM) as s:
+
-            s.bind(("", 0))
+        if self.subprocess:
-            s.setsockopt(socket.SOL_SOCKET, socket.SO_REUSEADDR, 1)
+            self.start_mlx_server()
-            return s.getsockname()[1]
+
+    def start_mlx_server(self):
+        if not os.getenv("MLX_PORT"):
+            self.port = self.find_free_port()
+            command = f"mlx_lm.server --model {self.model} --port {self.port}"
+            self.server_process = subprocess.Popen(command, shell=True)
+            logging.info(f"Started MLX server on port {self.port}")
+
+    def find_free_port(self):
+        import socket
+        s = socket.socket(socket.AF_INET, socket.SOCK_STREAM)
+        s.bind(('', 0))
+        port = s.getsockname()[1]
+        s.close()
+        return port
 
     async def on_startup(self):
-        # This function is called when the server is started.
+        logging.info(f"on_startup:{__name__}")
-        print(f"on_startup:{__name__}")
-        self.start_subprocess()
 
     async def on_shutdown(self):
-        # This function is called when the server is stopped.
+        if self.subprocess and hasattr(self, 'server_process'):
-        print(f"on_shutdown:{__name__}")
+            self.server_process.terminate()
-        self.stop_subprocess()
+            logging.info(f"Terminated MLX server on port {self.port}")
-
-    def start_subprocess(self):
-        # Start the subprocess for "mlx_lm.server --model ${MLX_MODEL} --port ${PORT}"
-        try:
-            self.process = subprocess.Popen(
-                ["mlx_lm.server", "--model", self.model, "--port", str(self.port)],
-                stdout=subprocess.PIPE,
-                stderr=subprocess.PIPE,
-            )
-            print(
-                f"Subprocess started with PID: {self.process.pid} on port {self.port}"
-            )
-        except Exception as e:
-            print(f"Failed to start subprocess: {e}")
-
-    def stop_subprocess(self):
-        # Stop the subprocess if it is running
-        if self.process:
-            try:
-                self.process.terminate()
-                self.process.wait()
-                print(f"Subprocess with PID {self.process.pid} terminated")
-            except Exception as e:
-                print(f"Failed to terminate subprocess: {e}")
 
     def get_response(
         self, user_message: str, model_id: str, messages: List[dict], body: dict
     ) -> Union[str, Generator, Iterator]:
-        # This is where you can add your custom pipelines like RAG.'
+        logging.info(f"get_response:{__name__}")
-        print(f"get_response:{__name__}")
 
-        MLX_BASE_URL = f"http://localhost:{self.port}"
+        url = f"http://{self.host}:{self.port}/v1/chat/completions"
-        MODEL = self.model
+        headers = {"Content-Type": "application/json"}
+
+        # Extract and validate parameters from the request body
+        max_tokens = body.get("max_tokens", 1024)
+        if not isinstance(max_tokens, int) or max_tokens < 0:
+            max_tokens = 1024  # Default to 1024 if invalid
 
-        # Extract additional parameters from the body
         temperature = body.get("temperature", 0.8)
-        max_tokens = body.get("max_tokens", 1000)
+        if not isinstance(temperature, (int, float)) or temperature < 0:
-        top_p = body.get("top_p", 1.0)
+            temperature = 0.8  # Default to 0.8 if invalid
-        repetition_penalty = body.get("repetition_penalty", 1.0)
-        stop = self.stop_sequences
 
-        if "user" in body:
+        repeat_penalty = body.get("repeat_penalty", 1.0)
-            print("######################################")
+        if not isinstance(repeat_penalty, (int, float)) or repeat_penalty < 0:
-            print(f'# User: {body["user"]["name"]} ({body["user"]["id"]})')
+            repeat_penalty = 1.0  # Default to 1.0 if invalid
-            print(f"# Message: {user_message}")
-            print("######################################")
 
         payload = {
-            "model": MODEL,
+            "messages": messages,
-            "messages": [message.model_dump() for message in messages],
-            "temperature": temperature,
             "max_tokens": max_tokens,
-            "top_p": top_p,
+            "temperature": temperature,
-            "repetition_penalty": repetition_penalty,
+            "repetition_penalty": repeat_penalty,
-            "stop": stop,
+            "stop": self.stop_sequence,
+            "stream": body.get("stream", False)
         }
 
         try:
-            r = requests.post(
+            r = requests.post(url, headers=headers, json=payload, stream=body.get("stream", False))
-                url=f"{MLX_BASE_URL}/v1/chat/completions",
-                json=payload,
-                stream=True,
-            )
-
             r.raise_for_status()
 
-            return r.iter_lines()
+            if body.get("stream", False):
+                return r.iter_lines()
+            else:
+                return r.json()
         except Exception as e:
             return f"Error: {e}"