Update app.py

app.py

@@ -22,7 +22,18 @@ def generate(
     
     prompt = pipe.tokenizer.apply_chat_template(message_template, tokenize=False, add_generation_prompt=True)
     
-    outputs = pipe(prompt, max_length=1024, do_sample=True, temperature=0.65, top_k=35, top_p=0.55, repetition_penalty=1.176)
+    if model_name == "Felladrin/Pythia-31M-Chat-v1":
+        outputs = pipe(prompt, max_new_tokens=250, use_cache=True, penalty_alpha=0.5, top_k=2, repetition_penalty=1.0016)
+    elif model_name == "Felladrin/Llama-68M-Chat-v1":
+        outputs = pipe(prompt, max_new_tokens=250, use_cache=True, penalty_alpha=0.5, top_k=4, repetition_penalty=1.043)
+    elif model_name == "Felladrin/Smol-Llama-101M-Chat-v1":
+        outputs = pipe(prompt, max_new_tokens=250, use_cache=True, penalty_alpha=0.5, top_k=4, repetition_penalty=1.105)
+    elif model_name == "Felladrin/Llama-160M-Chat-v1":
+        outputs = pipe(prompt, max_new_tokens=250, use_cache=True, penalty_alpha=0.5, top_k=4, repetition_penalty=1.01)
+    elif model_name == "Felladrin/TinyMistral-248M-SFT-v4":
+        outputs = pipe(prompt, max_new_tokens=250, use_cache=True, penalty_alpha=0.5, top_k=5, repetition_penalty=1.001)
+    else:
+        outputs = pipe(prompt, max_new_tokens=250, do_sample=True, temperature=0.65, top_k=35, top_p=0.55, repetition_penalty=1.176)
         
     return outputs[0]["generated_text"]