Upload 3 files

README.md

@@ -1,15 +1,33 @@
 ---
-title: Proteus2
-emoji: 🌖
+title: SaulLM Legal Assistant
+emoji: ⚖️
 colorFrom: blue
-colorTo: blue
+colorTo: indigo
 sdk: gradio
-sdk_version: 6.3.0
-python_version: '3.12'
+sdk_version: 5.9.1
 app_file: app.py
 pinned: false
 license: mit
-short_description: Hobby
 ---
 
-Check out the configuration reference at https://huggingface.co/docs/hub/spaces-config-reference
+# SaulLM-141B Legal Assistant
+
+Private legal AI assistant powered by Saul-141B, a specialized 141 billion parameter model for legal reasoning.
+
+## Features
+
+- **Specialized Legal Knowledge**: Trained on 540B legal tokens from U.S. and European legal systems
+- **Zero GPU**: Free 25 min/day of H200 compute time
+- **Private Queries**: Your legal questions stay private
+- **Chat Interface**: Natural conversation with legal context
+
+## Usage
+
+Ask about:
+- Statutory interpretation
+- Case law analysis
+- Legal concepts and definitions
+- Compliance questions
+- Contract review concepts
+
+**Note**: This provides informational analysis only, not legal advice. Consult qualified legal professionals for actual legal matters.

app.py

@@ -0,0 +1,122 @@
+import gradio as gr
+import spaces
+from transformers import AutoTokenizer, AutoModelForCausalLM
+import torch
+
+# Load model and tokenizer
+model_id = "Equall/SaulLM-141B-Instruct"
+tokenizer = AutoTokenizer.from_pretrained(model_id)
+model = AutoModelForCausalLM.from_pretrained(
+    model_id,
+    torch_dtype=torch.bfloat16,
+    device_map="auto"
+)
+
+@spaces.GPU()
+def generate_response(message, history, system_prompt, max_tokens, temperature):
+    """Generate legal analysis using Saul-141B"""
+    
+    # Build conversation history
+    messages = []
+    
+    if system_prompt:
+        messages.append({"role": "system", "content": system_prompt})
+    
+    for human, assistant in history:
+        messages.append({"role": "user", "content": human})
+        messages.append({"role": "assistant", "content": assistant})
+    
+    messages.append({"role": "user", "content": message})
+    
+    # Format for model
+    input_text = tokenizer.apply_chat_template(messages, tokenize=False, add_generation_prompt=True)
+    inputs = tokenizer(input_text, return_tensors="pt").to(model.device)
+    
+    # Generate
+    outputs = model.generate(
+        **inputs,
+        max_new_tokens=max_tokens,
+        temperature=temperature,
+        do_sample=temperature > 0,
+        pad_token_id=tokenizer.eos_token_id
+    )
+    
+    response = tokenizer.decode(outputs[0][inputs['input_ids'].shape[1]:], skip_special_tokens=True)
+    return response
+
+# Default system prompt for legal queries
+DEFAULT_SYSTEM = """You are SaulLM-141B, a specialized legal language model. You provide accurate legal analysis based on U.S. and European legal systems. 
+
+IMPORTANT DISCLAIMERS:
+- This is for informational purposes only, not legal advice
+- Information may not reflect recent legal developments
+- Users should consult qualified legal professionals for actual legal advice
+- Do not use this for decisions that could affect legal rights"""
+
+# Build interface
+with gr.Blocks(theme=gr.themes.Soft()) as demo:
+    gr.Markdown("# SaulLM-141B Legal Assistant")
+    gr.Markdown("*Specialized AI for legal reasoning and analysis. Private queries, powered by Zero GPU (25 min/day free).*")
+    
+    with gr.Row():
+        with gr.Column(scale=3):
+            chatbot = gr.Chatbot(label="Legal Analysis", height=500)
+            msg = gr.Textbox(
+                label="Your Legal Question",
+                placeholder="Ask about statutes, case law, legal concepts, or compliance...",
+                lines=3
+            )
+            with gr.Row():
+                submit = gr.Button("Submit", variant="primary")
+                clear = gr.Button("Clear Chat")
+        
+        with gr.Column(scale=1):
+            system_prompt = gr.Textbox(
+                label="System Prompt",
+                value=DEFAULT_SYSTEM,
+                lines=12,
+                max_lines=12
+            )
+            max_tokens = gr.Slider(
+                label="Max Response Tokens",
+                minimum=100,
+                maximum=2000,
+                value=1000,
+                step=100
+            )
+            temperature = gr.Slider(
+                label="Temperature",
+                minimum=0.0,
+                maximum=1.0,
+                value=0.7,
+                step=0.1
+            )
+            gr.Markdown("### Usage Tips")
+            gr.Markdown("""
+            - Be specific about jurisdiction
+            - Cite relevant statutes/cases if known
+            - Zero GPU resets after 60s idle
+            - 25 min/day free compute limit
+            """)
+    
+    def user_submit(message, history):
+        return "", history + [[message, None]]
+    
+    def bot_respond(history, system_prompt, max_tokens, temperature):
+        message = history[-1][0]
+        history_context = history[:-1]
+        
+        response = generate_response(message, history_context, system_prompt, max_tokens, temperature)
+        history[-1][1] = response
+        return history
+    
+    msg.submit(user_submit, [msg, chatbot], [msg, chatbot], queue=False).then(
+        bot_respond, [chatbot, system_prompt, max_tokens, temperature], chatbot
+    )
+    submit.click(user_submit, [msg, chatbot], [msg, chatbot], queue=False).then(
+        bot_respond, [chatbot, system_prompt, max_tokens, temperature], chatbot
+    )
+    clear.click(lambda: None, None, chatbot, queue=False)
+
+if __name__ == "__main__":
+    demo.queue().launch()

requirements.txt

@@ -0,0 +1,4 @@
+transformers
+torch
+spaces
+accelerate