use together

app.py

@@ -1,13 +1,12 @@
 import base64
 from pathlib import Path
 import streamlit as st
-from transformers import AutoModelForCausalLM, AutoTokenizer, pipeline
 from langchain.memory import ConversationBufferMemory
 from langchain.memory.chat_message_histories import StreamlitChatMessageHistory
 from langchain.chains import ConversationalRetrievalChain
 from langchain.embeddings import VoyageEmbeddings
-from langchain.vectorstores import SupabaseVectorStore
-from langchain.llms.huggingface_pipeline import HuggingFacePipeline
+from langchain.vectorstores.supabase import SupabaseVectorStore
+from langchain.llms.together import Together
 from st_supabase_connection import SupabaseConnection
 
 msgs = StreamlitChatMessageHistory()
@@ -35,32 +34,16 @@ def load_retriever():
     return vector_store.as_retriever()
 
 
-@st.cache_resource
-def load_model():
-    model_path = "llmware/bling-sheared-llama-1.3b-0.1"
-    tokenizer = AutoTokenizer.from_pretrained(model_path)
-    model = AutoModelForCausalLM.from_pretrained(
-        model_path,
-        offload_folder="offload",
-        offload_state_dict=True,
-        torch_dtype="auto",
-    ).eval()
-    pipe = pipeline(
-        "text-generation",
-        model=model,
-        tokenizer=tokenizer,
-        max_new_tokens=500,
-        eos_token_id=tokenizer.eos_token_id,
-        pad_token_id=tokenizer.eos_token_id,
-        do_sample=True,
-        temperature=0.3,
-    )
-    return HuggingFacePipeline(pipeline=pipe)
-
+llm = Together(
+    model="togethercomputer/StripedHyena-Nous-7B",
+    temperature=0.5,
+    max_tokens=200,
+    top_k=1,
+    together_api_key=st.secrets.together_api_key,
+)
 
-hf = load_model()
 retriever = load_retriever()
-chat = ConversationalRetrievalChain.from_llm(hf, retriever)
+chat = ConversationalRetrievalChain.from_llm(llm, retriever)
 
 st.markdown(
     "<div style='display: flex;justify-content: center;'><img width='150' src='data:image/png;base64,{}' class='img-fluid'></div>".format(