Upload 2 files

app.py

@@ -0,0 +1,72 @@
+import spaces
+import gradio as gr
+import torch
+from transformers import AutoModelForCausalLM, AutoTokenizer
+
+MODEL_ID = "devoppro/FastLLM"
+
+tokenizer = AutoTokenizer.from_pretrained(MODEL_ID)
+model = AutoModelForCausalLM.from_pretrained(
+    MODEL_ID,
+    torch_dtype=torch.float16,
+)
+model.to("cuda")
+model.eval()
+
+
+@spaces.GPU(duration=30)
+def generate(prompt, max_new_tokens, temperature, top_p):
+    inputs = tokenizer(prompt, return_tensors="pt").to("cuda")
+    with torch.no_grad():
+        output_ids = model.generate(
+            **inputs,
+            max_new_tokens=int(max_new_tokens),
+            do_sample=True,
+            temperature=float(temperature),
+            top_p=float(top_p),
+            pad_token_id=tokenizer.eos_token_id,
+        )
+    full_text = tokenizer.decode(output_ids[0], skip_special_tokens=True)
+    return full_text
+
+
+with gr.Blocks(title="FastLLM (150M) Demo") as demo:
+    gr.Markdown(
+        """
+        # FastLLM (150M) — Modern Causal Language Model
+        A ~150M parameter decoder-only model (GQA, SwiGLU, RMSNorm, RoPE) from
+        [devoppro/FastLLM](https://huggingface.co/devoppro/FastLLM).
+        Small model, so expect small-model quality — this is a demo, not a chatbot.
+        """
+    )
+    with gr.Row():
+        with gr.Column():
+            prompt = gr.Textbox(
+                label="Prompt",
+                value="Once upon a time,",
+                lines=4,
+            )
+            max_new_tokens = gr.Slider(16, 512, value=128, step=16, label="Max new tokens")
+            temperature = gr.Slider(0.1, 1.5, value=0.7, step=0.05, label="Temperature")
+            top_p = gr.Slider(0.1, 1.0, value=0.95, step=0.05, label="Top-p")
+            run_btn = gr.Button("Generate", variant="primary")
+        with gr.Column():
+            output = gr.Textbox(label="Output", lines=12)
+
+    run_btn.click(
+        fn=generate,
+        inputs=[prompt, max_new_tokens, temperature, top_p],
+        outputs=output,
+    )
+
+    gr.Examples(
+        examples=[
+            ["Once upon a time,", 128, 0.7, 0.95],
+            ["The most important thing about machine learning is", 128, 0.7, 0.95],
+            ["def fibonacci(n):", 96, 0.5, 0.9],
+        ],
+        inputs=[prompt, max_new_tokens, temperature, top_p],
+    )
+
+if __name__ == "__main__":
+    demo.queue().launch()

requirements .txt

@@ -0,0 +1,6 @@
+spaces
+gradio
+torch
+transformers
+accelerate
+safetensors