lemonade-sdk 8.0.0__tar.gz → 8.0.2__tar.gz

{lemonade_sdk-8.0.0/src/lemonade_sdk.egg-info → lemonade_sdk-8.0.2}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: lemonade-sdk
-Version: 8.0.0
+Version: 8.0.2
 Summary: Lemonade SDK: Your LLM Aide for Validation and Deployment
 Author-email: lemonade@amd.com
 Requires-Python: >=3.10, <3.12
@@ -133,7 +133,7 @@ Maximum LLM performance requires the right hardware accelerator with the right i
     <tr>
       <td>🎮 GPU</td>
       <td align="center">—</td>
-      <td align="center">Vulkan: All platforms<br><small>Focus: Radeon™ 7000/9000</small></td>
+      <td align="center">Vulkan: All platforms<br><small>Focus:<br/>Ryzen™ AI 7000/8000/300<br/>Radeon™ 7000/9000</small></td>
       <td align="center">—</td>
       <td align="center">✅</td>
       <td align="center">✅</td>
@@ -158,6 +158,10 @@ Maximum LLM performance requires the right hardware accelerator with the right i
 | **llamacpp** | Community-driven engine with strong GPU acceleration, support for thousands of `.gguf` models, and advanced features such as vision-language models (VLMs) and mixture-of-experts (MoEs). |
 | **Hugging Face (HF)** | Hugging Face's `transformers` library can run the original `.safetensors` trained weights for models on Meta's PyTorch engine, which provides a source of truth for accuracy measurement. |
 
+## Integrate Lemonade Server with Your Application
+
+Lemonade Server enables languages including Python, C++, Java, C#, Node.js, Go, Ruby, Rust, and PHP. For the full list and integration details, see [docs/server/README.md](./docs/server/README.md).
+
 ## Contributing
 
 We are actively seeking collaborators from across the industry. If you would like to contribute to this project, please check out our [contribution guide](./docs/contribute.md).

{lemonade_sdk-8.0.0 → lemonade_sdk-8.0.2}/README.md

@@ -55,7 +55,7 @@ Maximum LLM performance requires the right hardware accelerator with the right i
     <tr>
       <td>🎮 GPU</td>
       <td align="center">—</td>
-      <td align="center">Vulkan: All platforms<br><small>Focus: Radeon™ 7000/9000</small></td>
+      <td align="center">Vulkan: All platforms<br><small>Focus:<br/>Ryzen™ AI 7000/8000/300<br/>Radeon™ 7000/9000</small></td>
       <td align="center">—</td>
       <td align="center">✅</td>
       <td align="center">✅</td>
@@ -80,6 +80,10 @@ Maximum LLM performance requires the right hardware accelerator with the right i
 | **llamacpp** | Community-driven engine with strong GPU acceleration, support for thousands of `.gguf` models, and advanced features such as vision-language models (VLMs) and mixture-of-experts (MoEs). |
 | **Hugging Face (HF)** | Hugging Face's `transformers` library can run the original `.safetensors` trained weights for models on Meta's PyTorch engine, which provides a source of truth for accuracy measurement. |
 
+## Integrate Lemonade Server with Your Application
+
+Lemonade Server enables languages including Python, C++, Java, C#, Node.js, Go, Ruby, Rust, and PHP. For the full list and integration details, see [docs/server/README.md](./docs/server/README.md).
+
 ## Contributing
 
 We are actively seeking collaborators from across the industry. If you would like to contribute to this project, please check out our [contribution guide](./docs/contribute.md).

{lemonade_sdk-8.0.0 → lemonade_sdk-8.0.2}/src/lemonade/tools/report/table.py

@@ -7,6 +7,7 @@ from tabulate import tabulate
 import lemonade.common.build as build
 import lemonade.common.filesystem as fs
 from lemonade.cache import Keys
+from lemonade.tools.accuracy import LMEvalHarness
 from lemonade.tools.huggingface.bench import HuggingfaceBench
 from lemonade.tools.llamacpp.bench import LlamaCppBench
 from lemonade.tools.mmlu import AccuracyMMLU
@@ -527,6 +528,14 @@ class LemonadePerfTable(Table):
                     ".2f",
                 )
             ],
+            LMEvalHarness: [
+                AdditionalStat(
+                    "EleutherAI\nLM Evaluation",
+                    "^lm_eval_",
+                    "^lm_eval_",
+                    ".1f",
+                )
+            ],
         },
         "last_columns": [
             SimpleStat(

{lemonade_sdk-8.0.0 → lemonade_sdk-8.0.2}/src/lemonade/tools/server/llamacpp.py

@@ -416,25 +416,43 @@ def chat_completion(
         exclude_unset=True, exclude_none=True
     )
 
-    def event_stream():
+    # Check if streaming is requested
+    if chat_completion_request.stream:
+
+        def event_stream():
+            try:
+                # Enable streaming
+                for chunk in client.chat.completions.create(**request_dict):
+                    yield f"data: {chunk.model_dump_json()}\n\n"
+                yield "data: [DONE]\n\n"
+
+                # Show telemetry after completion
+                telemetry.show_telemetry()
+
+            except Exception as e:  # pylint: disable=broad-exception-caught
+                yield f'data: {{"error": "{str(e)}"}}\n\n'
+
+        return StreamingResponse(
+            event_stream(),
+            media_type="text/event-stream",
+            headers={
+                "Cache-Control": "no-cache",
+                "Connection": "keep-alive",
+            },
+        )
+    else:
+        # Non-streaming response
         try:
-            # Enable streaming
-            request_dict["stream"] = True
-            for chunk in client.chat.completions.create(**request_dict):
-                yield f"data: {chunk.model_dump_json()}\n\n"
-            yield "data: [DONE]\n\n"
+            # Disable streaming for non-streaming requests
+            response = client.chat.completions.create(**request_dict)
 
             # Show telemetry after completion
             telemetry.show_telemetry()
 
+            return response
+
         except Exception as e:  # pylint: disable=broad-exception-caught
-            yield f'data: {{"error": "{str(e)}"}}\n\n'
-
-    return StreamingResponse(
-        event_stream(),
-        media_type="text/event-stream",
-        headers={
-            "Cache-Control": "no-cache",
-            "Connection": "keep-alive",
-        },
-    )
+            raise HTTPException(
+                status_code=status.HTTP_500_INTERNAL_SERVER_ERROR,
+                detail=f"Chat completion error: {str(e)}",
+            )

{lemonade_sdk-8.0.0 → lemonade_sdk-8.0.2}/src/lemonade/tools/server/serve.py

@@ -29,6 +29,7 @@ from openai.types.chat.chat_completion_message_tool_call import (
     ChatCompletionMessageToolCall,
     Function,
 )
+from openai.types.completion_usage import CompletionUsage
 from openai.types.chat.chat_completion import Choice
 from openai.types.chat.chat_completion_chunk import (
     ChoiceDelta,
@@ -576,9 +577,16 @@ class Server(ManagementTool):
                 logprobs=logprobs,
             )
 
+            usage = CompletionUsage(
+                prompt_tokens=self.input_tokens,
+                completion_tokens=self.output_tokens,
+                total_tokens=self.input_tokens + self.output_tokens,
+            )
+
             return Completion(
                 id="0",
                 choices=[choice],
+                usage=usage,
                 model=self.llm_loaded.checkpoint,
                 object="text_completion",
                 created=int(time.time()),
@@ -773,9 +781,16 @@ class Server(ManagementTool):
                 logprobs=None,
             )
 
+            usage = CompletionUsage(
+                prompt_tokens=self.input_tokens,
+                completion_tokens=self.output_tokens,
+                total_tokens=self.input_tokens + self.output_tokens,
+            )
+
             return ChatCompletion(
                 id="0",
                 choices=[choice],
+                usage=usage,
                 model=self.llm_loaded.checkpoint,
                 object="chat.completion",
                 created=int(time.time()),

{lemonade_sdk-8.0.0 → lemonade_sdk-8.0.2}/src/lemonade/tools/server/tray.py

@@ -266,7 +266,7 @@ class LemonadeTray(SystemTray):
             self.logger.error(f"Error changing port: {str(e)}")
             self.show_balloon_notification("Error", f"Failed to change port: {str(e)}")
 
-    def upgrade_to_latest(self, icon, item):
+    def upgrade_to_latest(self, _, __):
         """
         Download and launch the Lemonade Server installer
         """
@@ -281,21 +281,34 @@ class LemonadeTray(SystemTray):
         installer_path = os.path.join(
             tempfile.gettempdir(), "Lemonade_Server_Installer.exe"
         )
+        if os.path.exists(installer_path):
+            os.remove(installer_path)
 
         # Download the installer
         response = requests.get(self.latest_version_url, stream=True)
         response.raise_for_status()
 
-        # Save the installer to disk
+        # Save the installer to disk and force write to disk
         with open(installer_path, "wb") as f:
             for chunk in response.iter_content(chunk_size=8192):
                 f.write(chunk)
+            f.flush()
+            os.fsync(f.fileno())
 
-        # Launch the installer
-        subprocess.Popen([installer_path], shell=True)
+        # Launch the installer as a completely detached process
+        # subprocess.DETACHED_PROCESS - Creates a process that's not attached to the console
+        # subprocess.CREATE_NEW_PROCESS_GROUP - Creates a new process group
+        # close_fds=True - Closes file descriptors to prevent inheritance
+        subprocess.Popen(
+            [installer_path],
+            creationflags=subprocess.DETACHED_PROCESS
+            | subprocess.CREATE_NEW_PROCESS_GROUP,
+            close_fds=True,
+            shell=True,
+            cwd=tempfile.gettempdir(),
+        )
 
-        # Quit the application
-        self.exit_app(icon, item)
+        # No need to quit the application, the installer will handle it
 
     def create_menu(self):
         """

lemonade_sdk-8.0.2/src/lemonade/version.py

@@ -0,0 +1 @@
+__version__ = "8.0.2"

{lemonade_sdk-8.0.0 → lemonade_sdk-8.0.2/src/lemonade_sdk.egg-info}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: lemonade-sdk
-Version: 8.0.0
+Version: 8.0.2
 Summary: Lemonade SDK: Your LLM Aide for Validation and Deployment
 Author-email: lemonade@amd.com
 Requires-Python: >=3.10, <3.12
@@ -133,7 +133,7 @@ Maximum LLM performance requires the right hardware accelerator with the right i
     <tr>
       <td>🎮 GPU</td>
       <td align="center">—</td>
-      <td align="center">Vulkan: All platforms<br><small>Focus: Radeon™ 7000/9000</small></td>
+      <td align="center">Vulkan: All platforms<br><small>Focus:<br/>Ryzen™ AI 7000/8000/300<br/>Radeon™ 7000/9000</small></td>
       <td align="center">—</td>
       <td align="center">✅</td>
       <td align="center">✅</td>
@@ -158,6 +158,10 @@ Maximum LLM performance requires the right hardware accelerator with the right i
 | **llamacpp** | Community-driven engine with strong GPU acceleration, support for thousands of `.gguf` models, and advanced features such as vision-language models (VLMs) and mixture-of-experts (MoEs). |
 | **Hugging Face (HF)** | Hugging Face's `transformers` library can run the original `.safetensors` trained weights for models on Meta's PyTorch engine, which provides a source of truth for accuracy measurement. |
 
+## Integrate Lemonade Server with Your Application
+
+Lemonade Server enables languages including Python, C++, Java, C#, Node.js, Go, Ruby, Rust, and PHP. For the full list and integration details, see [docs/server/README.md](./docs/server/README.md).
+
 ## Contributing
 
 We are actively seeking collaborators from across the industry. If you would like to contribute to this project, please check out our [contribution guide](./docs/contribute.md).

{lemonade_sdk-8.0.0 → lemonade_sdk-8.0.2}/src/lemonade_server/cli.py

@@ -4,7 +4,6 @@ import os
 from typing import Tuple, Optional
 import psutil
 from typing import List
-import subprocess
 
 
 # Error codes for different CLI scenarios
@@ -88,23 +87,26 @@ def stop():
         # Terminate the main process first
         process.terminate()
 
-        # Then terminate all children
+        # Then terminate llama-server child process (known to be stubborn)
+        # We avoid killing other child processes, such as the installer
         for child in children:
-            try:
-                child.terminate()
-            except psutil.NoSuchProcess:
-                pass  # Child already terminated
+            if "llama-server" in child.name():
+                try:
+                    child.terminate()
+                except psutil.NoSuchProcess:
+                    pass  # Child already terminated
 
         # Wait for main process
         process.wait(timeout=10)
 
-        # Kill any children that didn't terminate gracefully
+        # Kill llama-server child process if it didn't terminate gracefully
         for child in children:
-            try:
-                if child.is_running():
-                    child.kill()
-            except psutil.NoSuchProcess:
-                pass  # Child already terminated
+            if "llama-server" in child.name():
+                try:
+                    if child.is_running():
+                        child.kill()
+                except psutil.NoSuchProcess:
+                    pass  # Child already terminated
     except psutil.NoSuchProcess:
         # Process already terminated
         pass

{lemonade_sdk-8.0.0 → lemonade_sdk-8.0.2}/src/lemonade_server/pydantic_models.py

@@ -62,6 +62,7 @@ class ChatCompletionRequest(BaseModel):
     tools: list[dict] | None = None
     max_tokens: int | None = None
     max_completion_tokens: int | None = None
+    response_format: dict | None = None
 
 
 class ResponsesRequest(BaseModel):

lemonade_sdk-8.0.0/src/lemonade/version.py

@@ -1 +0,0 @@
-__version__ = "8.0.0"