Create rag.yml
Browse files
rag.yml
ADDED
|
@@ -0,0 +1,44 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
# TxtAI Rag Application
|
| 2 |
+
#
|
| 3 |
+
# All-in-one RAG application
|
| 4 |
+
#
|
| 5 |
+
# - Text Extraction with Docling, Chunking, and Indexing of documents. Supports PDF, DOCX, Web, XLSX files and more.
|
| 6 |
+
# - Vector database with embeddings generation
|
| 7 |
+
# - gpt-oss-20B LLM
|
| 8 |
+
# - RAG pipeline that joins vector search with the LLM
|
| 9 |
+
|
| 10 |
+
# Embeddings configuration
|
| 11 |
+
writable: True
|
| 12 |
+
embeddings:
|
| 13 |
+
content: True
|
| 14 |
+
|
| 15 |
+
# Text extraction
|
| 16 |
+
textractor:
|
| 17 |
+
sections: True
|
| 18 |
+
backend: docling
|
| 19 |
+
headers:
|
| 20 |
+
user-agent: Mozilla/5.0
|
| 21 |
+
minlength: 50
|
| 22 |
+
tuples: True
|
| 23 |
+
|
| 24 |
+
# RAG pipeline
|
| 25 |
+
rag:
|
| 26 |
+
path: unsloth/gpt-oss-20b-GGUF/gpt-oss-20b-Q4_K_M.gguf
|
| 27 |
+
n_ctx: 20000
|
| 28 |
+
system: You are a friendly assistant
|
| 29 |
+
output: flatten
|
| 30 |
+
template: |
|
| 31 |
+
Answer the following question using the provided context.
|
| 32 |
+
|
| 33 |
+
Question:
|
| 34 |
+
{question}
|
| 35 |
+
|
| 36 |
+
Context:
|
| 37 |
+
{context}
|
| 38 |
+
|
| 39 |
+
# Indexing workflow
|
| 40 |
+
workflow:
|
| 41 |
+
index:
|
| 42 |
+
tasks:
|
| 43 |
+
- textractor
|
| 44 |
+
- index
|