readme detais

README.md

@@ -9,4 +9,18 @@ app_file: app.py
 pinned: false
 ---
 
-An example chatbot using [Gradio](https://gradio.app), [`huggingface_hub`](https://huggingface.co/docs/huggingface_hub/v0.22.2/en/index), and the [Hugging Face Inference API](https://huggingface.co/docs/api-inference/index).
+An example chatbot using [Gradio](https://gradio.app), [`huggingface_hub`](https://huggingface.co/docs/huggingface_hub/v0.22.2/en/index), and the [Hugging Face Inference API](https://huggingface.co/docs/api-inference/index).
+
+### [HuggingFace Space with Quantized LLMs](https://huggingface.co/spaces/Robzy/llm)
+
+**Baseline model**: Llama-3.2-1B-Instruct with 4-bit quantization
+
+**Training infrastracture**:
+* Google Colab with NVIDIA Tesla T4 GPU
+* Finetuning with parameter-effecient finetuning (PEFT) by low-rank adaption (LORA) using Unsloth and HuggingFace's supervised finetuning libraries. 
+* Weight & Biases for model training monitoring and model checkpointing. Checkpointing every 10 steps.
+
+**Finetuning details**
+
+**Datasets**:
+* [Code instructions Alpaca 120k](https://huggingface.co/datasets/iamtarun/code_instructions_120k_alpaca)

app-old.py

@@ -1,78 +0,0 @@
-from llama_cpp import Llama
-import gradio as gr
-
-llm = Llama.from_pretrained(
-	repo_id="Robzy/Llama-3.2-1B-Instruct-Finetuned-q4_k_m",
-	filename="unsloth.Q4_K_M.gguf",
-)
-
-llm2 = Llama.from_pretrained(
-    repo_id="Robzy/Llama-3.2-1B-Instruct-Finetuned-16bit",
-    filename="unsloth.F16.gguf",
-)
-
-def predict(message, history):
-    messages = [{"role": "system", "content": "You are a helpful assistant."}]
-    for user_message, bot_message in history:
-        if user_message:
-            messages.append({"role": "user", "content": user_message})
-        if bot_message:
-            messages.append({"role": "assistant", "content": bot_message})
-    messages.append({"role": "user", "content": message})
-    
-    response = ""
-    for chunk in llm.create_chat_completion(
-        stream=True,
-        messages=messages,
-    ):
-        part = chunk["choices"][0]["delta"].get("content", None)
-        if part:
-            response += part
-        yield response
-
-
-def predict2(message, history):
-    messages = [{"role": "system", "content": "You are a helpful assistant."}]
-    for user_message, bot_message in history:
-        if user_message:
-            messages.append({"role": "user", "content": user_message})
-        if bot_message:
-            messages.append({"role": "assistant", "content": bot_message})
-    messages.append({"role": "user", "content": message})
-    
-    response = ""
-    for chunk in llm2.create_chat_completion(
-        stream=True,
-        messages=messages,
-    ):
-        part = chunk["choices"][0]["delta"].get("content", None)
-        if part:
-            response += part
-        yield response
-
-
-chat1 = gr.ChatInterface(predict, title="4-bit")
-chat2 = gr.ChatInterface(predict2, title="16-bit")
-chat3 = gr.ChatInterface(predict2, title="16-bit")
-
-def update_chat(value):
-    if value == "4-bit":
-        chat1.render(visible=True)
-        chat2.render(visible=False)
-        chat3.render(visible=False)
-    elif value == "16-bit":
-        chat1.render(visible=False)
-        chat2.render(visible=True)
-        chat3.render(visible=False)
-    else:
-        chat1.render(visible=False)
-        chat2.render(visible=False)
-        chat3.render(visible=True)
-
-with gr.Blocks() as demo:
-
-    gr.Markdown("# Quantized Llama Comparison for Code Generation")
-    dropdown = gr.Dropdown(["4-bit", "16-bit", "32-bit"], label="Choose model version", value="4-bit")
-    dropdown.change(fn=update_chat, inputs=dropdown, outputs=[chat1, chat2, chat3])
-
-demo.launch()