qtype 0.0.14__tar.gz → 0.0.16__tar.gz

{qtype-0.0.14/qtype.egg-info → qtype-0.0.16}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: qtype
-Version: 0.0.14
+Version: 0.0.16
 Summary: DSL for Generative AI Prototyping
 Author-email: Lou Kratz <lou.kratz+qtype@bazaarvoice.com>
 License-Expression: Apache-2.0

{qtype-0.0.14 → qtype-0.0.16}/pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "qtype"
-version = "0.0.14"
+version = "0.0.16"
 description = "DSL for Generative AI Prototyping"
 authors = [{ name="Lou Kratz", email="lou.kratz+qtype@bazaarvoice.com" }]
 readme = "README.md"

{qtype-0.0.14 → qtype-0.0.16}/qtype/interpreter/api.py

@@ -6,6 +6,7 @@ from pathlib import Path
 import pandas as pd
 from fastapi import FastAPI, HTTPException, Query
 from fastapi.middleware.cors import CORSMiddleware
+from fastapi.responses import RedirectResponse
 from fastapi.staticfiles import StaticFiles
 
 from qtype.dsl.base_types import StepCardinality
@@ -70,6 +71,7 @@ class APIExecutor:
                     StaticFiles(directory=str(ui_dir), html=True),
                     name="ui",
                 )
+                app.get("/")(lambda: RedirectResponse(url="/ui"))
 
         flows = self.definition.flows if self.definition.flows else []
 

{qtype-0.0.14 → qtype-0.0.16}/qtype/interpreter/steps/llm_inference.py

@@ -1,9 +1,16 @@
 import logging
 from typing import Any, Callable
 
+from llama_cloud import MessageRole as LlamaMessageRole
 from llama_index.core.base.llms.types import ChatResponse, CompletionResponse
 
-from qtype.dsl.domain_types import ChatMessage, Embedding
+from qtype.dsl.base_types import PrimitiveTypeEnum
+from qtype.dsl.domain_types import (
+    ChatContent,
+    ChatMessage,
+    Embedding,
+    MessageRole,
+)
 from qtype.interpreter.conversions import (
     from_chat_message,
     to_chat_message,
@@ -96,6 +103,20 @@ def execute(
             history: list[ChatMessage] = conversation_history
             inputs = [to_chat_message(msg) for msg in history] + inputs
 
+        if li.system_message and inputs[0].role != LlamaMessageRole.SYSTEM:
+            # There is a system prompt we should append
+            # Note system_prompt on the llm doesn't work for chat -- is only used for predict https://github.com/run-llama/llama_index/issues/13983
+            system_message = ChatMessage(
+                role=MessageRole.system,
+                blocks=[
+                    ChatContent(
+                        type=PrimitiveTypeEnum.text,
+                        content=li.system_message,
+                    )
+                ],
+            )
+            inputs = [to_chat_message(system_message)] + inputs
+
         # If the stream function is set, we'll stream the results
         chat_result: ChatResponse
         if stream_fn:

{qtype-0.0.14 → qtype-0.0.16/qtype.egg-info}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: qtype
-Version: 0.0.14
+Version: 0.0.16
 Summary: DSL for Generative AI Prototyping
 Author-email: Lou Kratz <lou.kratz+qtype@bazaarvoice.com>
 License-Expression: Apache-2.0