init files, idea

app.py

@@ -1,3 +1,30 @@
 """the idea is to embed all KBOB categories as vectores.
 then when a new document in added, we do a sim search with the doc vector in the KBOB vectores
-to map/classify. can be done in multiple steps. """
+to map/classify. can be done in multiple steps. """
+
+
+import streamlit as st
+from transformers import AutoModelForCausalLM, AutoTokenizer
+
+# Load the model and tokenizer
+model_name = "https://huggingface.co/mistralai/Mixtral-8x7B-Instruct-v0.1"
+model_name = "mistralai/Mixtral-8x7B-Instruct-v0.1" # gated
+model_name = "mistral-community/Mistral-7B-v0.2"
+tokenizer = AutoTokenizer.from_pretrained(model_name)
+model = AutoModelForCausalLM.from_pretrained(model_name)
+
+
+# Function to generate text based on a prompt
+def generate_text(llm_prompt):
+    inputs = tokenizer(llm_prompt, return_tensors="pt")
+    outputs = model.generate(inputs["input_ids"], max_length=100)
+    return tokenizer.decode(outputs[0], skip_special_tokens=True)
+
+
+# Set up the Streamlit interface
+st.title("Mistral 7B Text Generation App")
+prompt = st.text_area("Enter your prompt:", "Once upon a time")
+if st.button("Generate"):
+    with st.spinner('Generating...'):
+        generated_text = generate_text(prompt)
+        st.text_area("Generated Text:", generated_text, height=300)

requirements.txt

@@ -7,4 +7,7 @@ langchain~=0.1.16
 tiktoken~=0.6.0
 numpy~=1.26.4
 requests~=2.31.0
-faiss-cpu
+faiss-cpu
+transformers~=4.40.2
+pandas~=2.2.2
+pillow~=10.3.0