Upload 3 files

README.md

@@ -1,10 +1,10 @@
 ---
-title: Trurl-2-7b-pl-instruct GGML
-emoji: 🌍
-colorFrom: green
-colorTo: pink
+title: trurl-2-7b-pl-instruct_GGML
+emoji: 📚
+colorFrom: indigo
+colorTo: red
 sdk: gradio
-sdk_version: 3.40.1
+sdk_version: 3.38.0
 app_file: app.py
 pinned: false
 license: other

app.py

@@ -0,0 +1,54 @@
+import gradio as gr
+import random
+import time
+from ctransformers import AutoModelForCausalLM
+
+
+params = {
+        "max_new_tokens":512,
+        "stop":["<end>" ,"<|endoftext|>","["],
+        "temperature":0.7,
+        "top_p":0.8,
+        "stream":True,
+        "batch_size": 8}
+
+
+llm = AutoModelForCausalLM.from_pretrained("Aspik101/trurl-2-7b-pl-instruct_GGML", model_type="llama")
+
+with gr.Blocks() as demo:
+    chatbot = gr.Chatbot()
+    msg = gr.Textbox()
+    clear = gr.Button("Clear")
+
+    def user(user_message, history):
+        return "", history + [[user_message, None]]
+
+    def parse_history(hist):
+        history_ = ""
+        for q, a in hist:
+            history_ += f"<user>: {q } \n"
+            if a:
+                history_ += f"<assistant>: {a} \n"
+        return history_
+
+    def bot(history):
+        print("history: ",history)
+        prompt =  f"Jesteś AI assystentem. Odpowiadaj po polsku. {parse_history(history)}. <assistant>:"
+        print("prompt: ",prompt)
+        stream = llm(prompt, **params)
+        history[-1][1] = ""
+        answer_save = ""
+        for character in stream:
+            history[-1][1] += character
+            answer_save += character
+            time.sleep(0.005)
+            yield history
+
+        print("answer_save: ",answer_save)
+    msg.submit(user, [msg, chatbot], [msg, chatbot], queue=False).then(
+        bot, chatbot, chatbot
+    )
+    clear.click(lambda: None, None, chatbot, queue=False)
+     
+demo.queue()
+demo.launch()

requirements.txt

@@ -0,0 +1,4 @@
+torch
+ctransformers
+gradio
+dl-hf-model