Feature/pkgsplit (#83)

* Starting to spawn base package
* More package hacking
* Bedrock and VertexAI
* Parquet split
* Updated templates
* Utils

trustgraph-flow/trustgraph/model/text_completion/claude/llm.py

@@ -0,0 +1,199 @@
+
+"""
+Simple LLM service, performs text prompt completion using Claude.
+Input is prompt, output is response.
+"""
+
+import anthropic
+from prometheus_client import Histogram
+
+from .... schema import TextCompletionRequest, TextCompletionResponse, Error
+from .... schema import text_completion_request_queue
+from .... schema import text_completion_response_queue
+from .... log_level import LogLevel
+from .... base import ConsumerProducer
+from .... exceptions import TooManyRequests
+
+module = ".".join(__name__.split(".")[1:-1])
+
+default_input_queue = text_completion_request_queue
+default_output_queue = text_completion_response_queue
+default_subscriber = module
+default_model = 'claude-3-5-sonnet-20240620'
+default_temperature = 0.0
+default_max_output = 8192
+
+class Processor(ConsumerProducer):
+
+    def __init__(self, **params):
+    
+        input_queue = params.get("input_queue", default_input_queue)
+        output_queue = params.get("output_queue", default_output_queue)
+        subscriber = params.get("subscriber", default_subscriber)
+        model = params.get("model", default_model)
+        api_key = params.get("api_key")
+        temperature = params.get("temperature", default_temperature)
+        max_output = params.get("max_output", default_max_output)
+
+        super(Processor, self).__init__(
+            **params | {
+                "input_queue": input_queue,
+                "output_queue": output_queue,
+                "subscriber": subscriber,
+                "input_schema": TextCompletionRequest,
+                "output_schema": TextCompletionResponse,
+                "model": model,
+                "temperature": temperature,
+                "max_output": max_output,
+            }
+        )
+
+        if not hasattr(__class__, "text_completion_metric"):
+            __class__.text_completion_metric = Histogram(
+                'text_completion_duration',
+                'Text completion duration (seconds)',
+                buckets=[
+                    0.25, 0.5, 1.0, 2.0, 3.0, 4.0, 5.0, 6.0, 7.0,
+                    8.0, 9.0, 10.0, 11.0, 12.0, 13.0, 14.0, 15.0, 16.0,
+                    17.0, 18.0, 19.0, 20.0, 21.0, 22.0, 23.0, 24.0, 25.0,
+                    30.0, 35.0, 40.0, 45.0, 50.0, 60.0, 80.0, 100.0,
+                    120.0
+                ]
+            )
+
+        self.model = model
+        self.claude = anthropic.Anthropic(api_key=api_key)
+        self.temperature = temperature
+        self.max_output = max_output
+
+        print("Initialised", flush=True)
+
+    def handle(self, msg):
+
+        v = msg.value()
+
+        # Sender-produced ID
+
+        id = msg.properties()["id"]
+
+        print(f"Handling prompt {id}...", flush=True)
+
+        prompt = v.prompt
+
+        try:
+
+            # FIXME: Rate limits?
+
+            with __class__.text_completion_metric.time():
+
+                response = message = self.claude.messages.create(
+                    model=self.model,
+                    max_tokens=self.max_output,
+                    temperature=self.temperature,
+                    system = "You are a helpful chatbot.",
+                    messages=[
+                        {
+                            "role": "user",
+                            "content": [
+                                {
+                                    "type": "text",
+                                    "text": prompt
+                                }
+                            ]
+                        }
+                    ]
+                )
+
+            resp = response.content[0].text
+            inputtokens = response.usage.input_tokens
+            outputtokens = response.usage.output_tokens
+            print(resp, flush=True)
+            print(f"Input Tokens: {inputtokens}", flush=True)
+            print(f"Output Tokens: {outputtokens}", flush=True)
+
+            print("Send response...", flush=True)
+            r = TextCompletionResponse(response=resp, error=None, in_token=inputtokens, out_token=outputtokens, model=self.model)
+            self.send(r, properties={"id": id})
+
+            print("Done.", flush=True)
+
+        # FIXME: Wrong exception, don't know what this LLM throws
+        # for a rate limit
+        except TooManyRequests:
+
+            print("Send rate limit response...", flush=True)
+
+            r = TextCompletionResponse(
+                error=Error(
+                    type = "rate-limit",
+                    message = str(e),
+                ),
+                response=None,
+                in_token=None,
+                out_token=None,
+                model=None,
+            )
+
+            self.producer.send(r, properties={"id": id})
+
+            self.consumer.acknowledge(msg)
+
+        except Exception as e:
+
+            print(f"Exception: {e}")
+
+            print("Send error response...", flush=True)
+
+            r = TextCompletionResponse(
+                error=Error(
+                    type = "llm-error",
+                    message = str(e),
+                ),
+                response=None,
+                in_token=None,
+                out_token=None,
+                model=None,
+            )
+
+            self.producer.send(r, properties={"id": id})
+
+            self.consumer.acknowledge(msg)
+
+    @staticmethod
+    def add_args(parser):
+
+        ConsumerProducer.add_args(
+            parser, default_input_queue, default_subscriber,
+            default_output_queue,
+        )
+
+        parser.add_argument(
+            '-m', '--model',
+            default="claude-3-5-sonnet-20240620",
+            help=f'LLM model (default: claude-3-5-sonnet-20240620)'
+        )
+
+        parser.add_argument(
+            '-k', '--api-key',
+            help=f'Claude API key'
+        )
+
+        parser.add_argument(
+            '-t', '--temperature',
+            type=float,
+            default=default_temperature,
+            help=f'LLM temperature parameter (default: {default_temperature})'
+        )
+
+        parser.add_argument(
+            '-x', '--max-output',
+            type=int,
+            default=default_max_output,
+            help=f'LLM max output tokens (default: {default_max_output})'
+        )
+
+def run():
+
+    Processor.start(module, __doc__)
+
+