Update app.py from anycoder

app.py

@@ -2,9 +2,6 @@ import gradio as gr
 import numpy as np
 from typing import List, Tuple, Dict, Any
 import random
-import torch
-from transformers import AutoTokenizer, AutoModelForCausalLM
-import json
 
 class SolitaireEnvironment:
     def __init__(self):
@@ -22,7 +19,7 @@ class SolitaireEnvironment:
         # Deal cards to tableau (Solitaire rules)
         for i in range(7):
             self.tableau[i] = self.deck[:i+1]
-            self.deck = self.deck[i+1:]
+        self.deck = self.deck[i+1:]
         
     def get_valid_moves(self):
         # Simplified valid moves for demonstration
@@ -42,11 +39,7 @@ class SolitaireEnvironment:
 class SolitaireRLTrainer:
     def __init__(self):
         self.env = SolitaireEnvironment()
-        self.model_name = "mistralai/Mistral-7B-v0.1"  # Using a smaller model for demo
-        self.tokenizer = AutoTokenizer.from_pretrained(self.model_name)
-        if self.tokenizer.pad_token is None:
-            self.tokenizer.pad_token = self.tokenizer.eos_token
-            
+        
     def get_game_state(self):
         return {
             "tableau": self.env.tableau,
@@ -71,9 +64,8 @@ class MistralSolitaireAgent:
     def take_action(self, action: str):
         try:
             # Simulate game action and calculate reward
-        if "move" in action.lower():
             reward = random.uniform(0, 1)
-        return reward
+            return reward
 
 def train_mistral_solitaire(num_episodes: int, learning_rate: float):
     """Train Mistral model to play Solitaire using reinforcement learning"""
@@ -91,9 +83,9 @@ def train_mistral_solitaire(num_episodes: int, learning_rate: float):
 
 def play_solitaire_game(state_description: str, action: str):
     """Execute a move in the Solitaire game"""
-    # In a real implementation, this would modify the actual game state
+    # Simulate game action
     game_state = {
-        "tableau": [[random.randint(1, 13) for _ in range(random.randint(1, 5)] for _ in range(7)]
+        "tableau": [[random.randint(1, 13) for _ in range(i+1)] for i in range(7)]
     
     # Calculate reward based on action quality
     if "foundation" in action:
@@ -116,9 +108,12 @@ def format_game_state(state: Dict) -> str:
     
     # Tableau piles
     formatted += "### Tableau Piles\n"
-    for i, pile in enumerate(state.get("tableau", [])):
-        pile_str = " | ".join(str(card) for card in pile[-3:]]) if pile else "Empty"
-        formatted += "\n"
+    for i in range(7):
+        pile = state.get("tableau", [[]] * 7))[i]
+        if pile:
+            formatted += f"Pile {i+1}: {pile[-3:]} \n"
+        else:
+            formatted += f"Pile {i+1}: Empty\n"
         
     return formatted
 
@@ -126,13 +121,9 @@ def create_solitaire_ui():
     """Create the main Gradio interface for the Solitaire RL project"""
     
     with gr.Blocks() as demo:
-        gr.Markdown("# 🎮 Mistral 3B Solitaire RL Trainer")
-        gr.Markdown("Train Mistral 3B to play Solitaire using Reinforcement Learning")
+        gr.Markdown("# 🎮 Mistral Solitaire RL Trainer")
+        gr.Markdown("Train Mistral to play Solitaire using Reinforcement Learning")
         
-        with gr.Row():
-            with gr.Column(scale=1):
-                gr.Markdown("### 🏗️ Built with [anycoder](https://huggingface.co/spaces/akhaliq/anycoder)")
-                
         with gr.Tab("Training Interface"):
             with gr.Row():
                 episodes = gr.Slider(
@@ -140,16 +131,15 @@ def create_solitaire_ui():
                     minimum=10,
                     maximum=1000,
                     value=100,
-                    step=10,
-                    info="More episodes = better training but longer wait"
+                    step=10
                 )
                 learning_rate = gr.Slider(
                     label="Learning Rate",
                     minimum=0.001,
                     maximum=0.1,
                     value=0.01,
-                    step=0.001,
-                )
+                    step=0.001
+            )
                 
             train_btn = gr.Button("Start Training", variant="primary")
             training_output = gr.JSON(label="Training Progress")
@@ -165,16 +155,15 @@ def create_solitaire_ui():
             with gr.Row():
                 game_state = gr.Textbox(
                     label="Current Game State",
-                    value="A♠ 2♠ 3♠ | K♥ | Q♦ | J♣",
-                lines=3
+                    lines=3
             )
             
             with gr.Row():
                 action_input = gr.Textbox(
                     label="Action to Take",
                     placeholder="e.g., Move A♠ to foundation, Draw from deck"
-                )
-                
+            )
+            
             play_btn = gr.Button("Execute Move", variant="secondary")
             game_result = gr.JSON(label="Game Result")
             
@@ -193,15 +182,14 @@ def create_solitaire_ui():
             )
             
         with gr.Accordion("Advanced Options", open=False):
-            exploration_rate = gr.Slider(
-                label="Exploration Rate",
-                minimum=0.01,
+                exploration_rate = gr.Slider(
+                    label="Exploration Rate",
+                    minimum=0.01,
                     maximum=1.0,
                     value=0.1,
-                    step=0.01,
-                    info="Higher exploration = more experimentation"
-                )
-                
+                    step=0.01
+            )
+            
         gr.Markdown("---\n*This demo simulates training a language model to play Solitaire*")
         
     return demo
@@ -217,9 +205,6 @@ if __name__ == "__main__":
             text_size="lg",
             spacing_size="lg",
             radius_size="md"
-        ).set(
-            button_primary_background_fill="*primary_600",
-            button_primary_background_fill_hover="*primary_700"
         ),
-        footer_links=[{"label": "Built with anycoder", "url": "https://huggingface.co/spaces/akhaliq/anycoder"]
+        footer_links=[{"label": "Built with anycoder", "url": "https://huggingface.co/spaces/akhaliq/anycoder"
     )