feat: init app

.gitignore

@@ -0,0 +1,2 @@
+.aider*
+.env

app.py

@@ -1,64 +1,81 @@
+import json
 import gradio as gr
-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")
-
-
-def respond(
-    message,
-    history: list[tuple[str, str]],
-    system_message,
-    max_tokens,
-    temperature,
-    top_p,
-):
-    messages = [{"role": "system", "content": system_message}]
-
-    for val in history:
-        if val[0]:
-            messages.append({"role": "user", "content": val[0]})
-        if val[1]:
-            messages.append({"role": "assistant", "content": val[1]})
-
-    messages.append({"role": "user", "content": message})
-
-    response = ""
-
-    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
-
-        response += token
-        yield response
-
-
-"""
-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)",
-        ),
-    ],
+import torch
+from transformers import AutoModelForCausalLM, AutoTokenizer
+import spaces
+
+# Configuration
+MODEL_NAME = "speakleash/Bielik-11B-v2.3-Instruct"
+# DEVICE = "cuda" if torch.cuda.is_available() else "cpu"
+DEVICE = "cuda"
+TORCH_DTYPE = torch.bfloat16 if torch.cuda.is_available() else torch.float32
+MAX_TOKENS = 1000
+
+# Load model and tokenizer
+tokenizer = AutoTokenizer.from_pretrained(MODEL_NAME)
+model = AutoModelForCausalLM.from_pretrained(MODEL_NAME, torch_dtype=TORCH_DTYPE).to(
+    DEVICE
 )
 
+# Load prompts
+with open("prompts.json") as f:
+    prompts = json.load(f)
+
+
+@spaces.GPU
+def transform_text(prompt_name, user_input):
+    """Transform text using selected prompt and Bielik model"""
+    try:
+        # Get selected prompt
+        selected_prompt = next(p for p in prompts if p["name"] == prompt_name)
+
+        # Create messages structure
+        messages = [
+            {"role": "system", "content": selected_prompt["system_message"]},
+            {"role": "user", "content": user_input},
+        ]
+
+        # Tokenize and generate
+        input_ids = tokenizer.apply_chat_template(messages, return_tensors="pt").to(
+            DEVICE
+        )
+
+        generated_ids = model.generate(
+            input_ids, max_new_tokens=MAX_TOKENS, do_sample=True
+        )
+
+        return tokenizer.batch_decode(generated_ids)[0]
+
+    except Exception as e:
+        return f"Error: {str(e)}"
+
+
+# Create Gradio interface
+with gr.Blocks(title="Bielik Goblin") as interface:
+    gr.Markdown("# Bielik Goblin")
+
+    with gr.Row():
+        prompt_select = gr.Dropdown(
+            choices=[p["name"] for p in prompts],
+            label="Wybierz prompt",
+            interactive=True,
+        )
+
+    user_input = gr.Textbox(
+        label="Twój tekst", placeholder="Wpisz tutaj swój tekst...", lines=5
+    )
+
+    transform_btn = gr.Button("Przekształć tekst", variant="primary")
+
+    with gr.Column():
+        progress = gr.StatusTracker()
+        output = gr.Textbox(label="Wynik", interactive=False)
+
+    transform_btn.click(
+        fn=transform_text,
+        inputs=[prompt_select, user_input],
+        outputs=output,
+        status=progress,
+    )
 
-if __name__ == "__main__":
-    demo.launch()
+interface.queue().launch(debug=True)

prompts.json

@@ -0,0 +1,18 @@
+[
+    {
+        "name": "Parafraza",
+        "system_message": "Parafrazuj podany tekst zachowując jego główną ideę. Używaj naturalnego, potocznego języka polskiego."
+    },
+    {
+        "name": "Formalizacja",
+        "system_message": "Przekształć tekst na formalny styl biznesowy. Zachowaj wszystkie kluczowe informacje."
+    },
+    {
+        "name": "Korekta",
+        "system_message": "Popraw gramatykę, interpunkcję i styl tekstu. Zachowaj oryginalne znaczenie."
+    },
+    {
+        "name": "Podsumowanie",
+        "system_message": "Stwórz zwięzłe podsumowanie tekstu, zachowując kluczowe punkty. Maksymalnie 3 zdania."
+    }
+]

requirements.txt

@@ -1 +1,4 @@
-huggingface_hub==0.25.2
+gradio>=4.0
+transformers>=4.0
+torch>=2.0
+accelerate>=0.0