Update app.py

app.py

@@ -1,64 +1,162 @@
 import gradio as gr
+import torch
+import torch.nn as nn
+from torch.nn import functional as F
 from huggingface_hub import InferenceClient
 
-"""
-For more information on `huggingface_hub` Inference API support, please check the docs: https://huggingface.co/docs/huggingface_hub/v0.22.2/en/guides/inference
-"""
-client = InferenceClient("HuggingFaceH4/zephyr-7b-beta")
+# --- Model Definition (same as before) ---
+# NOTE: The model class MUST be defined in your app.py file
+# so that torch.load knows how to reconstruct it.
+# This code is a copy of the model classes from the training script.
+batch_size = 32
+block_size = 8
+n_embd = 32
+n_head = 4
+n_layer = 4
+dropout = 0.0
 
+class Head(nn.Module):
+    def __init__(self, head_size):
+        super().__init__()
+        self.key = nn.Linear(n_embd, head_size, bias=False)
+        self.query = nn.Linear(n_embd, head_size, bias=False)
+        self.value = nn.Linear(n_embd, head_size, bias=False)
+        self.register_buffer('tril', torch.tril(torch.ones(block_size, block_size)))
+        self.dropout = nn.Dropout(dropout)
+    def forward(self, x):
+        B, T, C = x.shape
+        k = self.key(x)
+        q = self.query(x)
+        wei = q @ k.transpose(-2, -1) * C**-0.5
+        wei = wei.masked_fill(self.tril[:T, :T] == 0, float('-inf'))
+        wei = F.softmax(wei, dim=-1)
+        self.dropout(wei)
+        v = self.value(x)
+        out = wei @ v
+        return out
 
-def respond(
-    message,
-    history: list[tuple[str, str]],
-    system_message,
-    max_tokens,
-    temperature,
-    top_p,
-):
-    messages = [{"role": "system", "content": system_message}]
+class MultiHeadAttention(nn.Module):
+    def __init__(self, num_heads, head_size):
+        super().__init__()
+        self.heads = nn.ModuleList([Head(head_size) for _ in range(num_heads)])
+        self.proj = nn.Linear(num_heads * head_size, n_embd)
+        self.dropout = nn.Dropout(dropout)
+    def forward(self, x):
+        out = torch.cat([h(x) for h in self.heads], dim=-1)
+        out = self.dropout(self.proj(out))
+        return out
 
-    for val in history:
-        if val[0]:
-            messages.append({"role": "user", "content": val[0]})
-        if val[1]:
-            messages.append({"role": "assistant", "content": val[1]})
+class FeedFoward(nn.Module):
+    def __init__(self, n_embd):
+        super().__init__()
+        self.net = nn.Sequential(
+            nn.Linear(n_embd, 4 * n_embd),
+            nn.ReLU(),
+            nn.Linear(4 * n_embd, n_embd),
+            nn.Dropout(dropout),
+        )
+    def forward(self, x):
+        return self.net(x)
 
-    messages.append({"role": "user", "content": message})
+class TransformerBlock(nn.Module):
+    def __init__(self, n_embd, n_head):
+        super().__init__()
+        head_size = n_embd // n_head
+        self.sa = MultiHeadAttention(n_head, head_size)
+        self.ffwd = FeedFoward(n_embd)
+        self.ln1 = nn.LayerNorm(n_embd)
+        self.ln2 = nn.LayerNorm(n_embd)
+    def forward(self, x):
+        x = x + self.sa(self.ln1(x))
+        x = x + self.ffwd(self.ln2(x))
+        return x
 
-    response = ""
+class LanguageModel(nn.Module):
+    def __init__(self, vocab_size, block_size, n_embd, n_head, n_layer, dropout):
+        super().__init__()
+        self.token_embedding_table = nn.Embedding(vocab_size, n_embd)
+        self.position_embedding_table = nn.Embedding(block_size, n_embd)
+        self.blocks = nn.Sequential(*[TransformerBlock(n_embd, n_head) for _ in range(n_layer)])
+        self.ln_f = nn.LayerNorm(n_embd)
+        self.lm_head = nn.Linear(n_embd, vocab_size)
+        self.block_size = block_size
+        self.vocab_size = vocab_size
 
-    for message in client.chat_completion(
-        messages,
-        max_tokens=max_tokens,
-        stream=True,
-        temperature=temperature,
-        top_p=top_p,
-    ):
-        token = message.choices[0].delta.content
+    def forward(self, idx, targets=None):
+        B, T = idx.shape
+        tok_emb = self.token_embedding_table(idx)
+        pos_emb = self.position_embedding_table(torch.arange(T, device=idx.device))
+        x = tok_emb + pos_emb
+        x = self.blocks(x)
+        x = self.ln_f(x)
+        logits = self.lm_head(x)
+        loss = None
+        if targets is not None:
+            B, T, C = logits.shape
+            logits = logits.view(B * T, C)
+            targets = targets.view(B * T)
+            loss = F.cross_entropy(logits, targets)
+        return logits, loss
 
-        response += token
-        yield response
+    def generate(self, idx, max_new_tokens):
+        for _ in range(max_new_tokens):
+            idx_cond = idx[:, -self.block_size:]
+            logits, loss = self(idx_cond)
+            logits = logits[:, -1, :]
+            probs = F.softmax(logits, dim=-1)
+            idx_next = torch.multinomial(probs, num_samples=1)
+            idx = torch.cat((idx, idx_next), dim=1)
+        return idx
 
 
+# --- Utility functions and model loading ---
+# We need to re-create the tokenizer and vocabulary from the training data.
+# In a real app, you would save and load these as well.
+# For simplicity, we are hardcoding a version of the dataset to get the vocab.
+jsonl_data = """
+{"text": "This is a much longer text that will serve as a simple dataset for our tiny language model. The model will learn to predict the next character based on the previous characters in the sequence."}
+{"text": "This demonstrates the core idea behind training an autoregressive language model. The quick brown fox jumps over the lazy dog."}
+{"text": "A journey of a thousand miles begins with a single step. The early bird catches the worm. All that glitters is not gold. A stitch in time saves nine."}
+{"text": "Where there's a will, there's a way. Look before you leap. You can't make an omelette without breaking a few eggs. Practice makes perfect. Don't count your chickens before they hatch."}
 """
-For information on how to customize the ChatInterface, peruse the gradio docs: https://www.gradio.app/docs/chatinterface
-"""
-demo = gr.ChatInterface(
-    respond,
-    additional_inputs=[
-        gr.Textbox(value="You are a friendly Chatbot.", label="System message"),
-        gr.Slider(minimum=1, maximum=2048, value=512, step=1, label="Max new tokens"),
-        gr.Slider(minimum=0.1, maximum=4.0, value=0.7, step=0.1, label="Temperature"),
-        gr.Slider(
-            minimum=0.1,
-            maximum=1.0,
-            value=0.95,
-            step=0.05,
-            label="Top-p (nucleus sampling)",
-        ),
+corpus = ""
+for line in jsonl_data.strip().split('\n'):
+    data_point = json.loads(line)
+    corpus += data_point['text']
+
+chars = sorted(list(set(corpus)))
+vocab_size = len(chars)
+stoi = {ch: i for i, ch in enumerate(chars)}
+itos = {i: ch for i, ch in enumerate(chars)}
+encode = lambda s: [stoi[c] for c in s]
+decode = lambda l: ''.join([itos[i] for i in l])
+device = 'cuda' if torch.cuda.is_available() else 'cpu'
+
+# Load the trained model.
+model = LanguageModel(vocab_size, block_size, n_embd, n_head, n_layer, dropout)
+model.load_state_dict(torch.load('model.pt', map_location=device))
+model.eval()  # Set the model to evaluation mode for inference.
+model.to(device)
+
+
+# --- Gradio UI & Inference function ---
+def generate_text(prompt, max_new_tokens):
+    # Encode the prompt text into tokens.
+    context = torch.tensor(encode(prompt), dtype=torch.long, device=device).unsqueeze(0)
+    # Generate new tokens.
+    generated_text_indices = model.generate(context, max_new_tokens=max_new_tokens)
+    # Decode the tokens back into text.
+    return decode(generated_text_indices[0].tolist())
+
+demo = gr.Interface(
+    fn=generate_text,
+    inputs=[
+        gr.Textbox(label="Prompt", placeholder="Enter your text prompt here..."),
+        gr.Slider(1, 100, value=20, step=1, label="Number of new tokens to generate"),
     ],
+    outputs="text",
+    title="Leaf Chat",
+    description="A demo application running on the Leaf model."
 )
 
-
-if __name__ == "__main__":
-    demo.launch()
+demo.launch()