Adding qwen llm model

examples/llm/.gitignore

@@ -0,0 +1 @@
+*.pt

examples/llm/qwen-dev.yml

@@ -0,0 +1,55 @@
+nodes:
+  - id: dora-microphone
+    build: pip install -e ../../node-hub/dora-microphone
+    path: dora-microphone
+    inputs:
+      tick: dora/timer/millis/2000
+    outputs:
+      - audio
+
+  - id: dora-vad
+    build: pip install -e ../../node-hub/dora-vad
+    path: dora-vad
+    inputs:
+      audio: dora-microphone/audio
+    outputs:
+      - audio
+
+  - id: dora-distil-whisper
+    build: pip install -e ../../node-hub/dora-distil-whisper
+    path: dora-distil-whisper
+    inputs:
+      input: dora-vad/audio
+    outputs:
+      - text
+    env:
+      TARGET_LANGUAGE: english
+
+  - id: dora-qwen
+    build: pip install -e ../../node-hub/dora-qwen
+    path: dora-qwen
+    inputs:
+      text: dora-distil-whisper/text
+    outputs:
+      - text
+
+  - id: plot
+    build: pip install -e ../../node-hub/dora-rerun
+    path: dora-rerun
+    inputs:
+      text_qwen: dora-qwen/text
+      text_whisper: dora-distil-whisper/text
+
+  - id: dora-outtetts
+    build: pip install -e ../../node-hub/dora-outtetts
+    path: dora-outtetts
+    inputs:
+      text: dora-qwen/text
+    outputs:
+      - audio
+
+  - id: dora-pyaudio
+    build: pip install -e ../../node-hub/dora-pyaudio
+    path: dora-pyaudio
+    inputs:
+      audio: dora-outtetts/audio

node-hub/dora-qwen/README.md

@@ -0,0 +1,37 @@
+# dora-qwen
+
+## Getting started
+
+- Install it with pip:
+
+```bash
+pip install -e .
+```
+
+## Contribution Guide
+
+- Format with [ruff](https://docs.astral.sh/ruff/):
+
+```bash
+ruff check . --fix
+```
+
+- Lint with ruff:
+
+```bash
+ruff check .
+```
+
+- Test with [pytest](https://github.com/pytest-dev/pytest)
+
+```bash
+pytest . # Test
+```
+
+## YAML Specification
+
+## Examples
+
+## License
+
+dora-qwen's code are released under the MIT License

node-hub/dora-qwen/dora_qwen/__init__.py

@@ -0,0 +1,11 @@
+import os
+
+# Define the path to the README file relative to the package directory
+readme_path = os.path.join(os.path.dirname(os.path.dirname(__file__)), "README.md")
+
+# Read the content of the README file
+try:
+    with open(readme_path, "r", encoding="utf-8") as f:
+        __doc__ = f.read()
+except FileNotFoundError:
+    __doc__ = "README file not found."

node-hub/dora-qwen/dora_qwen/__main__.py

@@ -0,0 +1,5 @@
+from .main import main
+
+
+if __name__ == "__main__":
+    main()

node-hub/dora-qwen/dora_qwen/main.py

@@ -0,0 +1,54 @@
+import os
+
+import pyarrow as pa
+from dora import Node
+from transformers import AutoModelForCausalLM, AutoTokenizer
+
+model_name = "Qwen/Qwen2.5-0.5B-Instruct"
+
+model = AutoModelForCausalLM.from_pretrained(
+    model_name, torch_dtype="auto", device_map="auto"
+)
+tokenizer = AutoTokenizer.from_pretrained(model_name)
+
+TRIGGER_WORDS = ["you", "wh", "tu"]
+
+
+def generate_response(prompt: str, history) -> str:
+    history += [{"role": "user", "content": prompt}]
+    text = tokenizer.apply_chat_template(
+        history, tokenize=False, add_generation_prompt=True
+    )
+    model_inputs = tokenizer([text], return_tensors="pt").to(model.device)
+    generated_ids = model.generate(**model_inputs, max_new_tokens=512)
+    generated_ids = [
+        output_ids[len(input_ids) :]
+        for input_ids, output_ids in zip(model_inputs.input_ids, generated_ids)
+    ]
+    response = tokenizer.batch_decode(generated_ids, skip_special_tokens=True)[0]
+    history += [{"role": "assistant", "content": response}]
+    return response, history
+
+
+def main():
+    history = [
+        {
+            "role": "system",
+            "content": "You are a Reachy robot, that gives extremely short answers only.",
+        },
+    ]
+    node = Node()
+
+    for event in node:
+        if event["type"] == "INPUT":
+            # Warning: Make sure to add my_output_id and my_input_id within the dataflow.
+            text = event["value"][0].as_py()
+            if any(word in text.lower() for word in TRIGGER_WORDS):
+                response, history = generate_response(text, history)
+                node.send_output(
+                    output_id="text", data=pa.array([response]), metadata={}
+                )
+
+
+if __name__ == "__main__":
+    main()

node-hub/dora-qwen/pyproject.toml

@@ -0,0 +1,16 @@
+[project]
+name = "dora-qwen"
+version = "0.0.0"
+authors = [{ name = "Your Name", email = "email@email.com" }]
+description = "dora-qwen"
+license = { text = "MIT" }
+readme = "README.md"
+requires-python = ">=3.8"
+
+dependencies = ["dora-rs >= 0.3.6"]
+
+[dependency-groups]
+dev = ["pytest >=8.1.1", "ruff >=0.9.1"]
+
+[project.scripts]
+dora-qwen = "dora_qwen.main:main"

node-hub/dora-qwen/tests/test_dora_qwen.py

@@ -0,0 +1,9 @@
+import pytest
+
+
+def test_import_main():
+    from dora_qwen.main import main
+
+    # Check that everything is working, and catch dora Runtime Exception as we're not running in a dora dataflow.
+    with pytest.raises(RuntimeError):
+        main()