github-actions commited on
Commit
128a507
·
1 Parent(s): 9c6d539

Auto deploy from GitHub Actions

Browse files
hf-space/hf-space/.github/workflows/deploy.yml ADDED
@@ -0,0 +1,43 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ name: Deploy to Hugging Face Spaces
2
+
3
+ on:
4
+ push:
5
+ branches:
6
+ - main
7
+
8
+ jobs:
9
+ deploy:
10
+ runs-on: ubuntu-latest
11
+
12
+ env:
13
+ FORCE_JAVASCRIPT_ACTIONS_TO_NODE24: true
14
+
15
+ steps:
16
+
17
+ - name: Checkout Repository
18
+ uses: actions/checkout@v5
19
+
20
+ - name: Deploy to Hugging Face Space
21
+
22
+ env:
23
+ HF_TOKEN: ${{ secrets.HF_TOKEN }}
24
+
25
+ run: |
26
+
27
+ git config --global user.email "github-actions@github.com"
28
+ git config --global user.name "github-actions"
29
+
30
+ git clone https://mdsalmon159:$HF_TOKEN@huggingface.co/spaces/mdsalmon159/hrchatbot hf-space
31
+
32
+ rsync -av \
33
+ --exclude='.git' \
34
+ --exclude='data/manual.pdf' \
35
+ ./ hf-space/
36
+
37
+ cd hf-space
38
+
39
+ git add .
40
+
41
+ git commit -m "Auto deploy from GitHub Actions" || echo "No changes to commit"
42
+
43
+ git push
hf-space/hf-space/.gitignore ADDED
@@ -0,0 +1,5 @@
 
 
 
 
 
 
1
+
2
+ __pycache__/
3
+ .ipynb_checkpoints/
4
+ *.pyc
5
+ data/manual.pdf
hf-space/hf-space/app.py ADDED
@@ -0,0 +1,45 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+
2
+ import gradio as gr
3
+
4
+ from rag_pipeline import ask_question
5
+
6
+ def respond(message, chat_history):
7
+
8
+ bot_message = ask_question(message)
9
+
10
+ chat_history.append(
11
+ (message, bot_message)
12
+ )
13
+
14
+ return "", chat_history
15
+
16
+ with gr.Blocks(theme=gr.themes.Soft()) as demo:
17
+
18
+ gr.Markdown(
19
+ "# Airlines HR Assistant"
20
+ )
21
+
22
+ chatbot = gr.Chatbot(
23
+ height=500
24
+ )
25
+
26
+ msg = gr.Textbox(
27
+ placeholder="Ask HR policy questions..."
28
+ )
29
+
30
+ clear = gr.Button("Clear Chat")
31
+
32
+ msg.submit(
33
+ respond,
34
+ [msg, chatbot],
35
+ [msg, chatbot]
36
+ )
37
+
38
+ clear.click(
39
+ lambda: None,
40
+ None,
41
+ chatbot,
42
+ queue=False
43
+ )
44
+
45
+ demo.launch()
hf-space/hf-space/config.py ADDED
@@ -0,0 +1,12 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+
2
+ MODEL_NAME = "sentence-transformers/all-MiniLM-L6-v2"
3
+
4
+ LLM_MODEL = "google/flan-t5-base"
5
+
6
+ PDF_PATH = "data/manual.pdf"
7
+
8
+ VECTOR_DB_PATH = "vectorstore/faiss_index"
9
+
10
+ CHUNK_SIZE = 500
11
+
12
+ CHUNK_OVERLAP = 50
hf-space/hf-space/hf-space/.gitattributes ADDED
@@ -0,0 +1,35 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ *.7z filter=lfs diff=lfs merge=lfs -text
2
+ *.arrow filter=lfs diff=lfs merge=lfs -text
3
+ *.bin filter=lfs diff=lfs merge=lfs -text
4
+ *.bz2 filter=lfs diff=lfs merge=lfs -text
5
+ *.ckpt filter=lfs diff=lfs merge=lfs -text
6
+ *.ftz filter=lfs diff=lfs merge=lfs -text
7
+ *.gz filter=lfs diff=lfs merge=lfs -text
8
+ *.h5 filter=lfs diff=lfs merge=lfs -text
9
+ *.joblib filter=lfs diff=lfs merge=lfs -text
10
+ *.lfs.* filter=lfs diff=lfs merge=lfs -text
11
+ *.mlmodel filter=lfs diff=lfs merge=lfs -text
12
+ *.model filter=lfs diff=lfs merge=lfs -text
13
+ *.msgpack filter=lfs diff=lfs merge=lfs -text
14
+ *.npy filter=lfs diff=lfs merge=lfs -text
15
+ *.npz filter=lfs diff=lfs merge=lfs -text
16
+ *.onnx filter=lfs diff=lfs merge=lfs -text
17
+ *.ot filter=lfs diff=lfs merge=lfs -text
18
+ *.parquet filter=lfs diff=lfs merge=lfs -text
19
+ *.pb filter=lfs diff=lfs merge=lfs -text
20
+ *.pickle filter=lfs diff=lfs merge=lfs -text
21
+ *.pkl filter=lfs diff=lfs merge=lfs -text
22
+ *.pt filter=lfs diff=lfs merge=lfs -text
23
+ *.pth filter=lfs diff=lfs merge=lfs -text
24
+ *.rar filter=lfs diff=lfs merge=lfs -text
25
+ *.safetensors filter=lfs diff=lfs merge=lfs -text
26
+ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
27
+ *.tar.* filter=lfs diff=lfs merge=lfs -text
28
+ *.tar filter=lfs diff=lfs merge=lfs -text
29
+ *.tflite filter=lfs diff=lfs merge=lfs -text
30
+ *.tgz filter=lfs diff=lfs merge=lfs -text
31
+ *.wasm filter=lfs diff=lfs merge=lfs -text
32
+ *.xz filter=lfs diff=lfs merge=lfs -text
33
+ *.zip filter=lfs diff=lfs merge=lfs -text
34
+ *.zst filter=lfs diff=lfs merge=lfs -text
35
+ *tfevents* filter=lfs diff=lfs merge=lfs -text
hf-space/hf-space/hf-space/README.md ADDED
@@ -0,0 +1,14 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ title: Hrchatbot
3
+ emoji: 📈
4
+ colorFrom: blue
5
+ colorTo: purple
6
+ sdk: gradio
7
+ sdk_version: 6.14.0
8
+ python_version: '3.13'
9
+ app_file: app.py
10
+ pinned: false
11
+ license: apache-2.0
12
+ ---
13
+
14
+ Check out the configuration reference at https://huggingface.co/docs/hub/spaces-config-reference
hf-space/hf-space/ingest.py ADDED
@@ -0,0 +1,34 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+
2
+ from langchain_community.document_loaders import PyPDFLoader
3
+
4
+ from langchain.text_splitter import RecursiveCharacterTextSplitter
5
+
6
+ from langchain_community.embeddings import HuggingFaceEmbeddings
7
+
8
+ from langchain_community.vectorstores import FAISS
9
+
10
+ from config import *
11
+
12
+ loader = PyPDFLoader(PDF_PATH)
13
+
14
+ documents = loader.load()
15
+
16
+ splitter = RecursiveCharacterTextSplitter(
17
+ chunk_size=500,
18
+ chunk_overlap=50
19
+ )
20
+
21
+ docs = splitter.split_documents(documents)
22
+
23
+ embeddings = HuggingFaceEmbeddings(
24
+ model_name="sentence-transformers/all-MiniLM-L6-v2"
25
+ )
26
+
27
+ vector_db = FAISS.from_documents(
28
+ docs,
29
+ embeddings
30
+ )
31
+
32
+ vector_db.save_local(VECTOR_DB_PATH)
33
+
34
+ print("Vector DB created successfully")
hf-space/hf-space/prompts.py ADDED
@@ -0,0 +1,60 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+
2
+ SYSTEM_PROMPT = """
3
+ You are an Airlines HR Assistant AI responsible for answering employee questions based strictly on the official HR handbook and company policy documents.
4
+
5
+ The user input will contain contextual information retrieved from the HR handbook.
6
+
7
+ The context section begins with:
8
+ ###Context
9
+
10
+ The employee question begins with:
11
+ ###Question
12
+
13
+ Your role is to provide accurate HR support using ONLY the provided handbook information.
14
+
15
+ Instructions:
16
+
17
+ - Answer ONLY using the information present in the provided context.
18
+ - Do NOT use external knowledge.
19
+ - Do NOT make assumptions or generate information not explicitly stated.
20
+ - Do NOT mention the context, document, handbook excerpts, or retrieval process in your response.
21
+ - Keep responses professional, concise, and employee-friendly.
22
+ - Use bullet points when appropriate for clarity.
23
+ - Maintain the tone of an HR representative.
24
+ - Focus strictly on HR and company policy topics including:
25
+ - Leave policies
26
+ - Attendance
27
+ - Payroll
28
+ - Employee conduct
29
+ - Benefits
30
+ - Working hours
31
+ - Travel policy
32
+ - Reimbursements
33
+ - Escalation procedures
34
+ - Employee responsibilities
35
+ - Company procedures
36
+
37
+ Rules for answering:
38
+
39
+ 1. If the answer exists in the context:
40
+ - Provide a direct and accurate answer.
41
+ - Keep formatting clean and readable.
42
+ - Avoid unnecessary elaboration.
43
+
44
+ 2. If the answer is partially available:
45
+ - Answer only the portion supported by the context.
46
+
47
+ 3. If the answer is NOT found in the context:
48
+ Respond EXACTLY with:
49
+ "I couldn't find this information in the handbook."
50
+
51
+ 4. Never fabricate policy details.
52
+
53
+ 5. Never answer questions unrelated to the HR handbook.
54
+
55
+ 6. If the user asks ethical, legal, medical, or safety-sensitive questions not covered in the handbook:
56
+ Respond EXACTLY with:
57
+ "I couldn't find this information in the handbook."
58
+
59
+ Answer carefully and ensure policy accuracy.
60
+ """
hf-space/hf-space/rag_pipeline.py ADDED
@@ -0,0 +1,75 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+
2
+ from transformers import pipeline
3
+
4
+ from langchain_community.llms import HuggingFacePipeline
5
+
6
+ from langchain.chains import RetrievalQA
7
+
8
+ from langchain.prompts import PromptTemplate
9
+
10
+ from langchain_community.embeddings import HuggingFaceEmbeddings
11
+
12
+ from langchain_community.vectorstores import FAISS
13
+
14
+ from config import *
15
+
16
+ from prompts import SYSTEM_PROMPT
17
+
18
+ template = """
19
+ {system_prompt}
20
+
21
+ ###Context:
22
+ {context}
23
+
24
+ ###Question:
25
+ {question}
26
+
27
+ Answer:
28
+ """
29
+
30
+ PROMPT = PromptTemplate(
31
+ template=template,
32
+ input_variables=["context", "question"],
33
+ partial_variables={
34
+ "system_prompt": SYSTEM_PROMPT
35
+ }
36
+ )
37
+
38
+ embeddings = HuggingFaceEmbeddings(
39
+ model_name="sentence-transformers/all-MiniLM-L6-v2"
40
+ )
41
+
42
+ vector_db = FAISS.load_local(
43
+ VECTOR_DB_PATH,
44
+ embeddings,
45
+ allow_dangerous_deserialization=True
46
+ )
47
+
48
+ retriever = vector_db.as_retriever(
49
+ search_kwargs={"k": 3}
50
+ )
51
+
52
+ pipe = pipeline(
53
+ "text2text-generation",
54
+ model="google/flan-t5-base",
55
+ max_new_tokens=256
56
+ )
57
+
58
+ llm = HuggingFacePipeline(
59
+ pipeline=pipe
60
+ )
61
+
62
+ qa_chain = RetrievalQA.from_chain_type(
63
+ llm=llm,
64
+ retriever=retriever,
65
+ chain_type="stuff",
66
+ chain_type_kwargs={
67
+ "prompt": PROMPT
68
+ }
69
+ )
70
+
71
+ def ask_question(query):
72
+
73
+ response = qa_chain.run(query)
74
+
75
+ return response
hf-space/hf-space/requirements.txt ADDED
@@ -0,0 +1,10 @@
 
 
 
 
 
 
 
 
 
 
 
1
+
2
+ gradio
3
+ langchain
4
+ langchain-community
5
+ sentence-transformers
6
+ transformers
7
+ accelerate
8
+ faiss-cpu
9
+ pypdf
10
+ torch
hf-space/hf-space/vectorstore/faiss_index/index.faiss ADDED
Binary file (73.8 kB). View file
 
hf-space/hf-space/vectorstore/faiss_index/index.pkl ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:5873301a1f6aab0cae850445e83a16e71f895000af15ff65d6dd28ed8d9873f4
3
+ size 26589
hf-space/rag_pipeline.py CHANGED
@@ -1,12 +1,5 @@
1
-
2
  from transformers import pipeline
3
 
4
- from langchain_community.llms import HuggingFacePipeline
5
-
6
- from langchain.chains import RetrievalQA
7
-
8
- from langchain.prompts import PromptTemplate
9
-
10
  from langchain_community.embeddings import HuggingFaceEmbeddings
11
 
12
  from langchain_community.vectorstores import FAISS
@@ -15,61 +8,53 @@ from config import *
15
 
16
  from prompts import SYSTEM_PROMPT
17
 
18
- template = """
19
- {system_prompt}
20
-
21
- ###Context:
22
- {context}
23
-
24
- ###Question:
25
- {question}
26
-
27
- Answer:
28
- """
29
-
30
- PROMPT = PromptTemplate(
31
- template=template,
32
- input_variables=["context", "question"],
33
- partial_variables={
34
- "system_prompt": SYSTEM_PROMPT
35
- }
36
- )
37
-
38
  embeddings = HuggingFaceEmbeddings(
39
- model_name="sentence-transformers/all-MiniLM-L6-v2"
40
  )
41
 
 
42
  vector_db = FAISS.load_local(
43
  VECTOR_DB_PATH,
44
  embeddings,
45
  allow_dangerous_deserialization=True
46
  )
47
 
 
48
  retriever = vector_db.as_retriever(
49
  search_kwargs={"k": 3}
50
  )
51
 
 
52
  pipe = pipeline(
53
  "text2text-generation",
54
- model="google/flan-t5-base",
55
  max_new_tokens=256
56
  )
57
 
58
- llm = HuggingFacePipeline(
59
- pipeline=pipe
60
- )
61
 
62
- qa_chain = RetrievalQA.from_chain_type(
63
- llm=llm,
64
- retriever=retriever,
65
- chain_type="stuff",
66
- chain_type_kwargs={
67
- "prompt": PROMPT
68
- }
69
- )
70
 
71
- def ask_question(query):
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
72
 
73
- response = qa_chain.run(query)
74
 
75
- return response
 
 
1
  from transformers import pipeline
2
 
 
 
 
 
 
 
3
  from langchain_community.embeddings import HuggingFaceEmbeddings
4
 
5
  from langchain_community.vectorstores import FAISS
 
8
 
9
  from prompts import SYSTEM_PROMPT
10
 
11
+ # Load embedding model
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
12
  embeddings = HuggingFaceEmbeddings(
13
+ model_name=MODEL_NAME
14
  )
15
 
16
+ # Load vector database
17
  vector_db = FAISS.load_local(
18
  VECTOR_DB_PATH,
19
  embeddings,
20
  allow_dangerous_deserialization=True
21
  )
22
 
23
+ # Create retriever
24
  retriever = vector_db.as_retriever(
25
  search_kwargs={"k": 3}
26
  )
27
 
28
+ # Load LLM
29
  pipe = pipeline(
30
  "text2text-generation",
31
+ model=LLM_MODEL,
32
  max_new_tokens=256
33
  )
34
 
35
+ def ask_question(query):
 
 
36
 
37
+ # Retrieve relevant docs
38
+ docs = retriever.get_relevant_documents(query)
 
 
 
 
 
 
39
 
40
+ context = "\n\n".join(
41
+ [doc.page_content for doc in docs]
42
+ )
43
+
44
+ prompt = f"""
45
+ {SYSTEM_PROMPT}
46
+
47
+ ###Context:
48
+ {context}
49
+
50
+ ###Question:
51
+ {query}
52
+
53
+ Answer:
54
+ """
55
+
56
+ result = pipe(prompt)
57
 
58
+ answer = result[0]["generated_text"]
59
 
60
+ return answer
rag_pipeline.py CHANGED
@@ -1,12 +1,5 @@
1
-
2
  from transformers import pipeline
3
 
4
- from langchain_community.llms import HuggingFacePipeline
5
-
6
- from langchain.chains import RetrievalQA
7
-
8
- from langchain.prompts import PromptTemplate
9
-
10
  from langchain_community.embeddings import HuggingFaceEmbeddings
11
 
12
  from langchain_community.vectorstores import FAISS
@@ -15,61 +8,53 @@ from config import *
15
 
16
  from prompts import SYSTEM_PROMPT
17
 
18
- template = """
19
- {system_prompt}
20
-
21
- ###Context:
22
- {context}
23
-
24
- ###Question:
25
- {question}
26
-
27
- Answer:
28
- """
29
-
30
- PROMPT = PromptTemplate(
31
- template=template,
32
- input_variables=["context", "question"],
33
- partial_variables={
34
- "system_prompt": SYSTEM_PROMPT
35
- }
36
- )
37
-
38
  embeddings = HuggingFaceEmbeddings(
39
- model_name="sentence-transformers/all-MiniLM-L6-v2"
40
  )
41
 
 
42
  vector_db = FAISS.load_local(
43
  VECTOR_DB_PATH,
44
  embeddings,
45
  allow_dangerous_deserialization=True
46
  )
47
 
 
48
  retriever = vector_db.as_retriever(
49
  search_kwargs={"k": 3}
50
  )
51
 
 
52
  pipe = pipeline(
53
  "text2text-generation",
54
- model="google/flan-t5-base",
55
  max_new_tokens=256
56
  )
57
 
58
- llm = HuggingFacePipeline(
59
- pipeline=pipe
60
- )
61
 
62
- qa_chain = RetrievalQA.from_chain_type(
63
- llm=llm,
64
- retriever=retriever,
65
- chain_type="stuff",
66
- chain_type_kwargs={
67
- "prompt": PROMPT
68
- }
69
- )
70
 
71
- def ask_question(query):
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
72
 
73
- response = qa_chain.run(query)
74
 
75
- return response
 
 
1
  from transformers import pipeline
2
 
 
 
 
 
 
 
3
  from langchain_community.embeddings import HuggingFaceEmbeddings
4
 
5
  from langchain_community.vectorstores import FAISS
 
8
 
9
  from prompts import SYSTEM_PROMPT
10
 
11
+ # Load embedding model
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
12
  embeddings = HuggingFaceEmbeddings(
13
+ model_name=MODEL_NAME
14
  )
15
 
16
+ # Load vector database
17
  vector_db = FAISS.load_local(
18
  VECTOR_DB_PATH,
19
  embeddings,
20
  allow_dangerous_deserialization=True
21
  )
22
 
23
+ # Create retriever
24
  retriever = vector_db.as_retriever(
25
  search_kwargs={"k": 3}
26
  )
27
 
28
+ # Load LLM
29
  pipe = pipeline(
30
  "text2text-generation",
31
+ model=LLM_MODEL,
32
  max_new_tokens=256
33
  )
34
 
35
+ def ask_question(query):
 
 
36
 
37
+ # Retrieve relevant docs
38
+ docs = retriever.get_relevant_documents(query)
 
 
 
 
 
 
39
 
40
+ context = "\n\n".join(
41
+ [doc.page_content for doc in docs]
42
+ )
43
+
44
+ prompt = f"""
45
+ {SYSTEM_PROMPT}
46
+
47
+ ###Context:
48
+ {context}
49
+
50
+ ###Question:
51
+ {query}
52
+
53
+ Answer:
54
+ """
55
+
56
+ result = pipe(prompt)
57
 
58
+ answer = result[0]["generated_text"]
59
 
60
+ return answer