trustgraph-flow 0.21.5__tar.gz → 0.21.6__tar.gz

{trustgraph-flow-0.21.5 → trustgraph-flow-0.21.6}/PKG-INFO

@@ -1,9 +1,9 @@
 Metadata-Version: 2.1
 Name: trustgraph-flow
-Version: 0.21.5
+Version: 0.21.6
 Summary: TrustGraph provides a means to run a pipeline of flexible AI processing components in a flexible means to achieve a processing pipeline.
 Home-page: https://github.com/trustgraph-ai/trustgraph
-Download-URL: https://github.com/trustgraph-ai/trustgraph/archive/refs/tags/v0.21.5.tar.gz
+Download-URL: https://github.com/trustgraph-ai/trustgraph/archive/refs/tags/v0.21.6.tar.gz
 Author: trustgraph.ai
 Author-email: security@trustgraph.ai
 Classifier: Programming Language :: Python :: 3

trustgraph-flow-0.21.6/scripts/text-completion-mistral

@@ -0,0 +1,6 @@
+#!/usr/bin/env python3
+
+from trustgraph.model.text_completion.mistral import run
+
+run()
+

{trustgraph-flow-0.21.5 → trustgraph-flow-0.21.6}/setup.py

@@ -50,6 +50,7 @@ setuptools.setup(
         "langchain-core",
         "langchain-text-splitters",
         "minio",
+        "mistralai"
         "neo4j",
         "ollama",
         "openai",
@@ -107,6 +108,7 @@ setuptools.setup(
         "scripts/text-completion-cohere",
         "scripts/text-completion-googleaistudio",
         "scripts/text-completion-llamafile",
+        "scripts/text-completion-mistral",
         "scripts/text-completion-ollama",
         "scripts/text-completion-openai",
         "scripts/triples-query-cassandra",

trustgraph-flow-0.21.6/trustgraph/flow_version.py

@@ -0,0 +1 @@
+__version__ = "0.21.6"

{trustgraph-flow-0.21.5 → trustgraph-flow-0.21.6}/trustgraph/gateway/document_embeddings_load.py

@@ -59,6 +59,6 @@ class DocumentEmbeddingsLoadEndpoint(SocketEndpoint):
                     ],
                 )
 
-                await self.publisher.send(None, elt)
+                self.publisher.send(None, elt)
 
         running.stop()

{trustgraph-flow-0.21.5 → trustgraph-flow-0.21.6}/trustgraph/gateway/graph_embeddings_load.py

@@ -36,6 +36,7 @@ class GraphEmbeddingsLoadEndpoint(SocketEndpoint):
     async def listener(self, ws, running):
         
         async for msg in ws:
+
             # On error, finish
             if msg.type == WSMsgType.ERROR:
                 break
@@ -59,7 +60,6 @@ class GraphEmbeddingsLoadEndpoint(SocketEndpoint):
                     ]
                 )
 
-                await self.publisher.send(None, elt)
-
+                self.publisher.send(None, elt)
 
         running.stop()

{trustgraph-flow-0.21.5 → trustgraph-flow-0.21.6}/trustgraph/gateway/requestor.py

@@ -53,11 +53,9 @@ class ServiceRequestor:
 
             q = self.sub.subscribe(id)
 
-            print("BOUT TO SEDN")
             await asyncio.to_thread(
                 self.pub.send, id, self.to_request(request)
             )
-            print("SENT")
 
             while True:
 

{trustgraph-flow-0.21.5 → trustgraph-flow-0.21.6}/trustgraph/gateway/socket.py

@@ -19,7 +19,7 @@ class SocketEndpoint:
         self.operation = "socket"
 
     async def listener(self, ws, running):
-        
+
         async for msg in ws:
             # On error, finish
             if msg.type == WSMsgType.TEXT:
@@ -53,7 +53,7 @@ class SocketEndpoint:
         try:
             await self.listener(ws, running)
         except Exception as e:
-            print(e, flush=True)
+            print("Socket exception:", e, flush=True)
 
         running.stop()
 

{trustgraph-flow-0.21.5 → trustgraph-flow-0.21.6}/trustgraph/gateway/triples_load.py

@@ -51,7 +51,7 @@ class TriplesLoadEndpoint(SocketEndpoint):
                     triples=to_subgraph(data["triples"]),
                 )
 
-                await self.publisher.send(None, elt)
+                self.publisher.send(None, elt)
 
 
         running.stop()

{trustgraph-flow-0.21.5 → trustgraph-flow-0.21.6}/trustgraph/model/text_completion/azure_openai/llm.py

@@ -23,9 +23,10 @@ default_output_queue = text_completion_response_queue
 default_subscriber = module
 default_temperature = 0.0
 default_max_output = 4192
-default_api = "2024-02-15-preview"
-default_endpoint = os.getenv("AZURE_ENDPOINT")
-default_token = os.getenv("AZURE_TOKEN")
+default_api = "2024-12-01-preview"
+default_endpoint = os.getenv("AZURE_ENDPOINT", None)
+default_token = os.getenv("AZURE_TOKEN", None)
+default_modeel = os.getenv("AZURE_MODEL", None)
 
 class Processor(ConsumerProducer):
 
@@ -34,12 +35,13 @@ class Processor(ConsumerProducer):
         input_queue = params.get("input_queue", default_input_queue)
         output_queue = params.get("output_queue", default_output_queue)
         subscriber = params.get("subscriber", default_subscriber)
-        endpoint = params.get("endpoint", default_endpoint)
-        token = params.get("token", default_token)
         temperature = params.get("temperature", default_temperature)
         max_output = params.get("max_output", default_max_output)
-        model = params.get("model")
+
         api = params.get("api_version", default_api)
+        endpoint = params.get("endpoint", default_endpoint)
+        token = params.get("token", default_token)
+        model = params.get("model", default_model)
 
         if endpoint is None:
             raise RuntimeError("Azure endpoint not specified")
@@ -177,6 +179,7 @@ class Processor(ConsumerProducer):
 
         parser.add_argument(
             '-e', '--endpoint',
+            default=default_endpoint,
             help=f'LLM model endpoint'
         )
 
@@ -188,11 +191,13 @@ class Processor(ConsumerProducer):
 
         parser.add_argument(
             '-k', '--token',
+            default=default_token,
             help=f'LLM model token'
         )
 
         parser.add_argument(
             '-m', '--model',
+            default=default_model,
             help=f'LLM model'
         )
 

trustgraph-flow-0.21.6/trustgraph/model/text_completion/mistral/llm.py

@@ -0,0 +1,201 @@
+
+"""
+Simple LLM service, performs text prompt completion using Mistral.
+Input is prompt, output is response.
+"""
+
+from mistralai import Mistral, RateLimitError
+from prometheus_client import Histogram
+import os
+
+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 = 'ministral-8b-latest'
+default_temperature = 0.0
+default_max_output = 4096
+default_api_key = os.getenv("MISTRAL_TOKEN")
+
+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", default_api_key)
+        temperature = params.get("temperature", default_temperature)
+        max_output = params.get("max_output", default_max_output)
+
+        if api_key is None:
+            raise RuntimeError("Mistral API key not specified")
+
+        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.temperature = temperature
+        self.max_output = max_output
+        self.mistral = Mistral(api_key=api_key)
+
+        print("Initialised", flush=True)
+
+    async def handle(self, msg):
+
+        v = msg.value()
+
+        # Sender-produced ID
+
+        id = msg.properties()["id"]
+
+        print(f"Handling prompt {id}...", flush=True)
+
+        prompt = v.system + "\n\n" + v.prompt
+
+        try:
+
+            with __class__.text_completion_metric.time():
+
+                resp = self.mistral.chat.complete(
+                    model=self.model,
+                    messages=[
+                        {
+                            "role": "user",
+                            "content": [
+                                {
+                                    "type": "text",
+                                    "text": prompt
+                                }
+                            ]
+                        }
+                    ],
+                    temperature=self.temperature,
+                    max_tokens=self.max_output,
+                    top_p=1,
+                    frequency_penalty=0,
+                    presence_penalty=0,
+                    response_format={
+                        "type": "text"
+                    }
+                )
+            
+            inputtokens = resp.usage.prompt_tokens
+            outputtokens = resp.usage.completion_tokens
+            print(resp.choices[0].message.content, 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.choices[0].message.content,
+                error=None,
+                in_token=inputtokens,
+                out_token=outputtokens,
+                model=self.model
+            )
+            await 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 Mistral.RateLimitError:
+
+            # Leave rate limit retries to the base handler
+            raise TooManyRequests()
+
+        except Exception as e:
+
+            # Apart from rate limits, treat all exceptions as unrecoverable
+
+            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,
+            )
+
+            await self.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=default_model,
+            help=f'LLM model (default: ministral-8b-latest)'
+        )
+
+        parser.add_argument(
+            '-k', '--api-key',
+            default=default_api_key,
+            help=f'Mistral 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.launch(module, __doc__)
+
+    

trustgraph-flow-0.21.6/trustgraph/model/text_completion/openai/__init__.py

@@ -0,0 +1,3 @@
+
+from . llm import *
+

trustgraph-flow-0.21.6/trustgraph/model/text_completion/openai/__main__.py

@@ -0,0 +1,7 @@
+#!/usr/bin/env python3
+
+from . llm import run
+
+if __name__ == '__main__':
+    run()
+

{trustgraph-flow-0.21.5 → trustgraph-flow-0.21.6}/trustgraph_flow.egg-info/PKG-INFO

@@ -1,9 +1,9 @@
 Metadata-Version: 2.1
 Name: trustgraph-flow
-Version: 0.21.5
+Version: 0.21.6
 Summary: TrustGraph provides a means to run a pipeline of flexible AI processing components in a flexible means to achieve a processing pipeline.
 Home-page: https://github.com/trustgraph-ai/trustgraph
-Download-URL: https://github.com/trustgraph-ai/trustgraph/archive/refs/tags/v0.21.5.tar.gz
+Download-URL: https://github.com/trustgraph-ai/trustgraph/archive/refs/tags/v0.21.6.tar.gz
 Author: trustgraph.ai
 Author-email: security@trustgraph.ai
 Classifier: Programming Language :: Python :: 3

{trustgraph-flow-0.21.5 → trustgraph-flow-0.21.6}/trustgraph_flow.egg-info/SOURCES.txt

@@ -40,6 +40,7 @@ scripts/text-completion-claude
 scripts/text-completion-cohere
 scripts/text-completion-googleaistudio
 scripts/text-completion-llamafile
+scripts/text-completion-mistral
 scripts/text-completion-ollama
 scripts/text-completion-openai
 scripts/triples-query-cassandra
@@ -181,6 +182,9 @@ trustgraph/model/text_completion/googleaistudio/llm.py
 trustgraph/model/text_completion/llamafile/__init__.py
 trustgraph/model/text_completion/llamafile/__main__.py
 trustgraph/model/text_completion/llamafile/llm.py
+trustgraph/model/text_completion/mistral/__init__.py
+trustgraph/model/text_completion/mistral/__main__.py
+trustgraph/model/text_completion/mistral/llm.py
 trustgraph/model/text_completion/ollama/__init__.py
 trustgraph/model/text_completion/ollama/__main__.py
 trustgraph/model/text_completion/ollama/llm.py

{trustgraph-flow-0.21.5 → trustgraph-flow-0.21.6}/trustgraph_flow.egg-info/requires.txt

@@ -13,7 +13,7 @@ langchain-community
 langchain-core
 langchain-text-splitters
 minio
-neo4j
+mistralaineo4j
 ollama
 openai
 pinecone[grpc]

trustgraph-flow-0.21.5/trustgraph/flow_version.py

@@ -1 +0,0 @@
-__version__ = "0.21.5"