Update app.py

app.py

@@ -13,7 +13,7 @@ if 'together_web3' not in st.session_state:
     st.session_state.together_web3 = TogetherWeb3()
 if 'loop' not in st.session_state:
     st.session_state.loop = asyncio.new_event_loop()
-async def _inference(prompt, max_tokens):
+async def _inference(prompt, max_tokens, stop, top_p, temperature, seed):
     result = await st.session_state.together_web3.language_model_inference(
         from_dict(
             data_class=LanguageModelInferenceRequest,
@@ -21,6 +21,10 @@ async def _inference(prompt, max_tokens):
                 "model": "Together-gpt-JT-6B-v1",
                 "max_tokens": max_tokens,
                 "prompt": prompt,
+                "stop": stop,
+                "top_p": top_p,
+                "temperature": temperature,
+                "seed": seed,
             }
         ),
     )
@@ -30,13 +34,14 @@ async def _inference(prompt, max_tokens):
 def infer(prompt, 
           model_name, 
           max_new_tokens=10, 
-          temperature=0.0,
+          temperature=1.0,
           top_p=1.0,
           num_completions=1,
           seed=42,
           stop="\n"):
     print("prompt", prompt)
-    response = st.session_state.loop.run_until_complete(_inference(prompt, int(max_new_tokens)))
+    stop = stop.split(";")
+    response = st.session_state.loop.run_until_complete(_inference(prompt, int(max_new_tokens), stop, float(top_p), float(temperature), int(seed)))
     print(response)
     return response.choices[0].text
     
@@ -45,7 +50,7 @@ col1, col2 = st.columns([1, 3])
 with col1:
     model_name = st.selectbox("Model", ["GPT-JT-6B-v1"])
     max_new_tokens = st.text_input('Max new tokens', "10")
-    temperature = st.text_input('temperature', "0.0")
+    temperature = st.text_input('temperature', "1.0")
     top_p = st.text_input('top_p', "1.0")
     num_completions = st.text_input('num_completions (only the best one will be returend)', "1")
     stop = st.text_input('stop, split by;', r'\n')