Fix WIP

app.py

@@ -11,7 +11,7 @@ with gr.Blocks() as iface:
         """# Petals playground
         **Let's play with prompts and inference settings for BLOOM and BLOOMZ 176B models!**
 
-        This space uses websocket API of [chat.petals.ml](http://chat.petals.ml). Health status of Petals network [lives here](http://health.petals.ml).
+        This space uses websocket API of [chat.petals.dev](http://chat.petals.dev). Health status of Petals network [lives here](http://health.petals.dev).
 
         Do NOT talk to BLOOM as an entity, it's not a chatbot but a webpage/blog/article completion model.
         For the best results: MIMIC a few sentences of a webpage similar to the content you want to generate.

chat.py

@@ -6,8 +6,8 @@ import gradio as gr
 
 import chat_client
 
-CHAT_URL = "ws://chat.petals.ml/api/v2/generate"
-# CHAT_URL='ws://localhost:8000/api/v2/generate'
+CHAT_URL = "ws://chat.petals.dev/api/v2/generate"
+#CHAT_URL='ws://localhost:8000/api/v2/generate'
 
 EMPTY_STATE = {
     "generate": False,
@@ -81,7 +81,7 @@ def _generate(
 
         try:
             state["client"] = chat_client.ModelClient(CHAT_URL)
-            state["client"].open_session(f"bigscience/{model}-petals", max_length)
+            state["client"].open_session(model, max_length)
             state["model"] = model
         except Exception as e:
             print(datetime.now(), str(e)[-500:])
@@ -211,7 +211,7 @@ with gr.Blocks() as iface_chat:
 
     with gr.Row():
         model = gr.Radio(
-            ["bloom", "bloomz", "bloom-7b1"], value="bloomz", label="Use model"
+            ["petals-team/StableBeluga2", "meta-llama/Llama-2-70b-chat-hf", "bigscience/bloomz"], value="petals-team/StableBeluga2", label="Use model"
         )
 
         # Additional ending sequence, at which generation shoud stop

chat_client.py

@@ -59,8 +59,8 @@ class ModelClient(object):
  
 def main():
     client = ModelClient("ws://localhost:8000/api/v2/generate")
-    # client = ModelClient("ws://chat.petals.ml/api/v2/generate")
-    client.open_session("bigscience/bloom-petals", 128)
+    #client = ModelClient("ws://chat.petals.dev/api/v2/generate")
+    client.open_session("stabilityai/StableBeluga2", 128)
 
     if len(sys.argv) > 1:
         prompt = sys.argv[1]

prompt.py

@@ -4,8 +4,8 @@ import gradio as gr
 
 import chat_client
 
-CHAT_URL = "ws://chat.petals.ml/api/v2/generate"
-# CHAT_URL='ws://localhost:8000/api/v2/generate'
+CHAT_URL = "ws://chat.petals.dev/api/v2/generate"
+#CHAT_URL='ws://localhost:8000/api/v2/generate'
 
 
 def generate(state, *args):
@@ -48,7 +48,7 @@ def _generate(
 
     try:
         client = chat_client.ModelClient(CHAT_URL)
-        client.open_session(f"bigscience/{model}-petals", max_length)
+        client.open_session(model, max_length)
     except Exception as e:
         print(datetime.now(), str(e)[-500:])
         raise gr.Error(str(e)[-500:])
@@ -94,7 +94,7 @@ def _generate(
             temperature=temperature,
             top_k=top_k,
             top_p=top_p,
-            extra_stop_sequences=seq,
+            stop_sequences=seq,
         ):
 
             if not state["generate"]:
@@ -137,7 +137,7 @@ with gr.Blocks() as iface_prompt:
 
     with gr.Row():
         model = gr.Radio(
-            ["bloom", "bloomz", "bloom-7b1"], value="bloom", label="Use model"
+            ["petals-team/StableBeluga2", "codellama/CodeLlama-34b-Instruct-hf", " meta-llama/Llama-2-70b-chat-hf", "meta-llama/Llama-2-70b-hf", "bigscience/bloomz", "bigscience/bloom"], value="stabilityai/StableBeluga2", label="Use model"
         )
 
         # Additional ending sequence, at which generation shoud stop
@@ -258,7 +258,7 @@ with gr.Blocks() as iface_prompt:
             ],
             [
                 "Lorem ipsum dolor sit amet, ",
-                "bloom",
+                "petals-team/StableBeluga2",
                 True,
                 0,
                 0.9,