llama-cpp-haystack 0.4.2__tar.gz → 0.4.3__tar.gz

llama_cpp_haystack-0.4.3/CHANGELOG.md

@@ -0,0 +1,93 @@
+# Changelog
+
+## [integrations/llama_cpp-v0.4.2] - 2024-12-10
+
+### 🧪 Testing
+
+- Do not retry tests in `hatch run test` command (#954)
+
+### ⚙️ CI
+
+- Adopt uv as installer (#1142)
+
+### 🧹 Chores
+
+- Update ruff linting scripts and settings (#1105)
+- Unpin `llama-cpp-python` (#1115)
+- Fix linting/isort (#1215)
+- Use text instead of content for ChatMessage in Llama.cpp, Langfuse and Mistral (#1238)
+
+### 🌀 Miscellaneous
+
+- Chore: lamma_cpp - ruff update, don't ruff tests (#998)
+- Fix: pin `llama-cpp-python<0.3.0` (#1111)
+
+## [integrations/llama_cpp-v0.4.1] - 2024-08-08
+
+### 🐛 Bug Fixes
+
+- Replace DynamicChatPromptBuilder with ChatPromptBuilder (#940)
+
+### ⚙️ CI
+
+- Retry tests to reduce flakyness (#836)
+
+### 🧹 Chores
+
+- Update ruff invocation to include check parameter (#853)
+- Pin `llama-cpp-python>=0.2.87` (#955)
+
+### 🌀 Miscellaneous
+
+- Ci: install `pytest-rerunfailures` where needed; add retry config to `test-cov` script (#845)
+- Fix: pin llama-cpp-python to an older version (#943)
+- Refactor: introduce `_convert_message_to_llamacpp_format` utility function (#939)
+
+## [integrations/llama_cpp-v0.4.0] - 2024-05-13
+
+### 🐛 Bug Fixes
+
+- Llama.cpp: change wrong links and imports (#436)
+- Fix order of API docs (#447)
+
+### 📚 Documentation
+
+- Update category slug (#442)
+- Small consistency improvements (#536)
+- Disable-class-def (#556)
+
+### 🧹 Chores
+
+- [**breaking**] Rename model_path to model in the Llama.cpp integration (#243)
+
+### 🌀 Miscellaneous
+
+- Generate api docs (#353)
+- Model_name_or_path > model (#418)
+- Llama.cpp - review docstrings (#510)
+- Llama.cpp - update examples (#511)
+- Make tests show coverage (#566)
+- Remove references to Python 3.7 (#601)
+- Chore: add license classifiers (#680)
+- Chore: change the pydoc renderer class (#718)
+- Basic implementation of llama.cpp chat generation (#723)
+
+## [integrations/llama_cpp-v0.2.1] - 2024-01-18
+
+### 🌀 Miscellaneous
+
+- Update import paths for beta5 (#233)
+
+## [integrations/llama_cpp-v0.2.0] - 2024-01-17
+
+### 🌀 Miscellaneous
+
+- Mount llama_cpp in haystack_integrations (#217)
+
+## [integrations/llama_cpp-v0.1.0] - 2024-01-09
+
+### 🚀 Features
+
+- Add Llama.cpp Generator (#179)
+
+<!-- generated by git-cliff -->

{llama_cpp_haystack-0.4.2 → llama_cpp_haystack-0.4.3}/PKG-INFO

@@ -1,13 +1,14 @@
-Metadata-Version: 2.3
+Metadata-Version: 2.4
 Name: llama-cpp-haystack
-Version: 0.4.2
+Version: 0.4.3
 Summary: An integration between the llama.cpp LLM framework and Haystack
 Project-URL: Documentation, https://github.com/deepset-ai/haystack-core-integrations/tree/main/integrations/llama_cpp#readme
 Project-URL: Issues, https://github.com/deepset-ai/haystack-core-integrations/issues
 Project-URL: Source, https://github.com/deepset-ai/haystack-core-integrations/tree/main/integrations/llama_cpp
 Author: Ashwin Mathur
 Author-email: deepset GmbH <info@deepset.ai>
-License: Apache-2.0
+License-Expression: Apache-2.0
+License-File: LICENSE.txt
 Classifier: Development Status :: 4 - Beta
 Classifier: License :: OSI Approved :: Apache Software License
 Classifier: Programming Language :: Python

{llama_cpp_haystack-0.4.2 → llama_cpp_haystack-0.4.3}/src/haystack_integrations/components/generators/llama_cpp/chat/chat_generator.py

@@ -2,7 +2,7 @@ import logging
 from typing import Any, Dict, List, Optional
 
 from haystack import component
-from haystack.dataclasses import ChatMessage, ChatRole
+from haystack.dataclasses import ChatMessage
 from llama_cpp import Llama
 from llama_cpp.llama_tokenizer import LlamaHFTokenizer
 
@@ -21,6 +21,10 @@ def _convert_message_to_llamacpp_format(message: ChatMessage) -> Dict[str, str]:
     if message.name:
         formatted_msg["name"] = message.name
 
+    if formatted_msg["role"] == "tool":
+        formatted_msg["name"] = message.tool_call_result.origin.tool_name
+        formatted_msg["content"] = message.tool_call_result.result
+
     return formatted_msg
 
 
@@ -114,26 +118,31 @@ class LlamaCppChatGenerator:
         formatted_messages = [_convert_message_to_llamacpp_format(msg) for msg in messages]
 
         response = self.model.create_chat_completion(messages=formatted_messages, **updated_generation_kwargs)
-        replies = [
-            ChatMessage(
-                content=choice["message"]["content"],
-                role=ChatRole[choice["message"]["role"].upper()],
-                name=None,
-                meta={
-                    "response_id": response["id"],
-                    "model": response["model"],
-                    "created": response["created"],
-                    "index": choice["index"],
-                    "finish_reason": choice["finish_reason"],
-                    "usage": response["usage"],
-                },
-            )
-            for choice in response["choices"]
-        ]
-
-        for reply, choice in zip(replies, response["choices"]):
+
+        replies = []
+
+        for choice in response["choices"]:
+            meta = {
+                "response_id": response["id"],
+                "model": response["model"],
+                "created": response["created"],
+                "index": choice["index"],
+                "finish_reason": choice["finish_reason"],
+                "usage": response["usage"],
+            }
+
+            name = None
             tool_calls = choice.get("message", {}).get("tool_calls", [])
             if tool_calls:
-                reply.meta["tool_calls"] = tool_calls
-        reply.name = tool_calls[0]["function"]["name"] if tool_calls else None
+                meta["tool_calls"] = tool_calls
+                name = tool_calls[0]["function"]["name"]
+
+            reply = ChatMessage.from_assistant(choice["message"]["content"], meta=meta)
+            if name:
+                if hasattr(reply, "_name"):
+                    reply._name = name  # new ChatMessage
+                elif hasattr(reply, "name"):
+                    reply.name = name  # legacy ChatMessage
+            replies.append(reply)
+
         return {"replies": replies}

{llama_cpp_haystack-0.4.2 → llama_cpp_haystack-0.4.3}/tests/test_chat_generator.py

@@ -41,11 +41,11 @@ def test_convert_message_to_llamacpp_format():
     assert _convert_message_to_llamacpp_format(message) == {"role": "user", "content": "I have a question"}
 
     message = ChatMessage.from_function("Function call", "function_name")
-    assert _convert_message_to_llamacpp_format(message) == {
-        "role": "function",
-        "content": "Function call",
-        "name": "function_name",
-    }
+    converted_message = _convert_message_to_llamacpp_format(message)
+
+    assert converted_message["role"] in ("function", "tool")
+    assert converted_message["name"] == "function_name"
+    assert converted_message["content"] == "Function call"
 
 
 class TestLlamaCppChatGenerator:

llama_cpp_haystack-0.4.2/CHANGELOG.md

@@ -1,50 +0,0 @@
-# Changelog
-
-## [integrations/llama_cpp-v0.4.1] - 2024-08-08
-
-### 🐛 Bug Fixes
-
-- Replace DynamicChatPromptBuilder with ChatPromptBuilder (#940)
-
-### ⚙️ Miscellaneous Tasks
-
-- Retry tests to reduce flakyness (#836)
-- Update ruff invocation to include check parameter (#853)
-- Pin `llama-cpp-python>=0.2.87` (#955)
-
-## [integrations/llama_cpp-v0.4.0] - 2024-05-13
-
-### 🐛 Bug Fixes
-
-- Fix commit (#436)
-
-
-- Fix order of API docs (#447)
-
-This PR will also push the docs to Readme
-
-### 📚 Documentation
-
-- Update category slug (#442)
-- Small consistency improvements (#536)
-- Disable-class-def (#556)
-
-### ⚙️ Miscellaneous Tasks
-
-- [**breaking**] Rename model_path to model in the Llama.cpp integration (#243)
-
-### Llama.cpp
-
-- Generate api docs (#353)
-
-## [integrations/llama_cpp-v0.2.1] - 2024-01-18
-
-## [integrations/llama_cpp-v0.2.0] - 2024-01-17
-
-## [integrations/llama_cpp-v0.1.0] - 2024-01-09
-
-### 🚀 Features
-
-- Add Llama.cpp Generator (#179)
-
-<!-- generated by git-cliff -->