new model

.gitignore

@@ -0,0 +1,25 @@
+# Python
+__pycache__/
+*.py[cod]
+*.pyo
+*.pyd
+
+# OS-filer
+.DS_Store
+Thumbs.db
+
+# Hugging Face cache eller stora modeller
+*.bin
+*.pt
+*.onnx
+*.safetensors
+
+# Loggar och tillfälliga filer
+*.log
+*.zip
+*.tmp
+*.cache/
+
+# Lokala miljöinställningar
+.env
+.venv

app.py

@@ -1,7 +1,7 @@
 import gradio as gr
 from langchain_community.document_loaders import PyPDFLoader
 from langchain_community.vectorstores import FAISS
-from langchain_community.embeddings import HuggingFaceEmbeddings
+from langchain_huggingface.embeddings import HuggingFaceEmbeddings
 from langchain_community.llms import HuggingFacePipeline
 from langchain.chains import RetrievalQA
 from transformers import pipeline
@@ -10,9 +10,9 @@ import os
 # 1. Ladda och indexera alla PDF:er i mappen "dokument/"
 def load_vectorstore():
     all_docs = []
-    for filename in os.listdir("document"):
+    for filename in os.listdir("dokument"):
         if filename.endswith(".pdf"):
-            path = os.path.join("document", filename)
+            path = os.path.join("dokument", filename)
             loader = PyPDFLoader(path)
             docs = loader.load_and_split()
             all_docs.extend(docs)
@@ -21,12 +21,13 @@ def load_vectorstore():
 
 vectorstore = load_vectorstore()
 
-# 2. Initiera Zephyr-modellen
-def load_zephyr():
-    pipe = pipeline("text-generation", model="HuggingFaceH4/zephyr-7b-beta")
+# 2. Initiera en mindre modell
+
+def load_model():
+    pipe = pipeline("text-generation", model="google/gemma-2b-it")
     return HuggingFacePipeline(pipeline=pipe, model_kwargs={"temperature": 0.3, "max_new_tokens": 512})
 
-llm = load_zephyr() 
+llm = load_model()
 
 # 3. Bygg QA-kedjan
 qa_chain = RetrievalQA.from_chain_type(llm=llm, retriever=vectorstore.as_retriever())

requirements.txt

@@ -2,6 +2,7 @@ huggingface_hub==0.25.2
 gradio
 langchain[all]>=0.1.14
 langchain-community>=0.0.19
+langchain-huggingface>=0.0.6
 transformers
 sentence-transformers
 faiss-cpu