Spaces:
Runtime error
Runtime error
bek-cs-request
#3
by bek-cs - opened
- app.py +8 -8
- vector_db/chunks.pkl +2 -2
- vector_db/index.faiss +2 -2
- vector_db/metadata.pkl +2 -2
app.py
CHANGED
|
@@ -17,7 +17,7 @@ with open("vector_db/metadata.pkl", "rb") as f:
|
|
| 17 |
ST = SentenceTransformer("BAAI/bge-large-en-v1.5")
|
| 18 |
github_base_url = "https://github.com/arsiba/EDPB-AI/blob/main/"
|
| 19 |
|
| 20 |
-
model_id = "
|
| 21 |
bnb = BitsAndBytesConfig(
|
| 22 |
load_in_4bit=True,
|
| 23 |
bnb_4bit_use_double_quant=True,
|
|
@@ -40,7 +40,6 @@ SYS = (
|
|
| 40 |
"Answer this Question:"
|
| 41 |
)
|
| 42 |
|
| 43 |
-
@spaces.GPU()
|
| 44 |
def retrieve(q, k=3):
|
| 45 |
emb = ST.encode(q)
|
| 46 |
D, I = index.search(np.array([emb], dtype="float32"), k)
|
|
@@ -78,8 +77,8 @@ def build_markdown_chunks(docs):
|
|
| 78 |
return "\n\n".join(lines)
|
| 79 |
|
| 80 |
@spaces.GPU()
|
| 81 |
-
def qa_fn(
|
| 82 |
-
docs, file_sources = retrieve(
|
| 83 |
file_links = build_markdown_links(file_sources)
|
| 84 |
markdown_chunks = build_markdown_chunks(docs)
|
| 85 |
prompt = make_prompt(question, docs)[:8000]
|
|
@@ -102,6 +101,8 @@ def qa_fn(faiss_search, question, top_k, temperature, max_tokens):
|
|
| 102 |
output = output.split("Output:", 1)[1].strip()
|
| 103 |
return "\n# Generated Answer\n", output,"\n# Used Documents\n", file_links, "\n# Used Context\n", markdown_chunks
|
| 104 |
|
|
|
|
|
|
|
| 105 |
heading_answer = gr.Markdown(label="Answer Heading")
|
| 106 |
outputs_answer = gr.Textbox(label="Answer")
|
| 107 |
heading_links = gr.Markdown(label="Links Heading")
|
|
@@ -112,11 +113,10 @@ outputs_chunks = gr.Markdown(label="Used Chunks")
|
|
| 112 |
demo = gr.Interface(
|
| 113 |
fn=qa_fn,
|
| 114 |
inputs=[
|
| 115 |
-
gr.Textbox(lines=
|
| 116 |
-
gr.Textbox(lines=1, label="What is your question?", placeholder="Please change to get propper results:\nWhat does the EDPB recommend regarding automated decision-making and profiling under the GDPR, and what rights do individuals have in relation to such decisions?"),
|
| 117 |
],
|
| 118 |
additional_inputs=[
|
| 119 |
-
gr.Slider(1,
|
| 120 |
gr.Slider(0.1, 1.0, value=0.6, step=0.05, label="Temperature"),
|
| 121 |
gr.Slider(64, 1024, value=512, step=64, label="Max Answer Length")
|
| 122 |
],
|
|
@@ -136,4 +136,4 @@ demo = gr.Interface(
|
|
| 136 |
)
|
| 137 |
|
| 138 |
if __name__ == "__main__":
|
| 139 |
-
demo.launch(share=True)
|
|
|
|
| 17 |
ST = SentenceTransformer("BAAI/bge-large-en-v1.5")
|
| 18 |
github_base_url = "https://github.com/arsiba/EDPB-AI/blob/main/"
|
| 19 |
|
| 20 |
+
model_id = "nvidia/Llama-3.1-Nemotron-Nano-8B-v1"
|
| 21 |
bnb = BitsAndBytesConfig(
|
| 22 |
load_in_4bit=True,
|
| 23 |
bnb_4bit_use_double_quant=True,
|
|
|
|
| 40 |
"Answer this Question:"
|
| 41 |
)
|
| 42 |
|
|
|
|
| 43 |
def retrieve(q, k=3):
|
| 44 |
emb = ST.encode(q)
|
| 45 |
D, I = index.search(np.array([emb], dtype="float32"), k)
|
|
|
|
| 77 |
return "\n\n".join(lines)
|
| 78 |
|
| 79 |
@spaces.GPU()
|
| 80 |
+
def qa_fn(question, top_k, temperature, max_tokens):
|
| 81 |
+
docs, file_sources = retrieve(question, top_k)
|
| 82 |
file_links = build_markdown_links(file_sources)
|
| 83 |
markdown_chunks = build_markdown_chunks(docs)
|
| 84 |
prompt = make_prompt(question, docs)[:8000]
|
|
|
|
| 101 |
output = output.split("Output:", 1)[1].strip()
|
| 102 |
return "\n# Generated Answer\n", output,"\n# Used Documents\n", file_links, "\n# Used Context\n", markdown_chunks
|
| 103 |
|
| 104 |
+
# TODO first model to make prompt more efficient
|
| 105 |
+
|
| 106 |
heading_answer = gr.Markdown(label="Answer Heading")
|
| 107 |
outputs_answer = gr.Textbox(label="Answer")
|
| 108 |
heading_links = gr.Markdown(label="Links Heading")
|
|
|
|
| 113 |
demo = gr.Interface(
|
| 114 |
fn=qa_fn,
|
| 115 |
inputs=[
|
| 116 |
+
gr.Textbox(lines=2, label="Your Question"),
|
|
|
|
| 117 |
],
|
| 118 |
additional_inputs=[
|
| 119 |
+
gr.Slider(1, 7, value=4, step=1, label="Top-K Documents"),
|
| 120 |
gr.Slider(0.1, 1.0, value=0.6, step=0.05, label="Temperature"),
|
| 121 |
gr.Slider(64, 1024, value=512, step=64, label="Max Answer Length")
|
| 122 |
],
|
|
|
|
| 136 |
)
|
| 137 |
|
| 138 |
if __name__ == "__main__":
|
| 139 |
+
demo.launch(share=True)
|
vector_db/chunks.pkl
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
-
size
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:5cd4fbdc599d57400b805d94897976dcd4c86584623dcd48a1eab3767cfb4e45
|
| 3 |
+
size 35695836
|
vector_db/index.faiss
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
-
size
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:05a2605f128bb07d1c9c780e2dc23ddb4ed961eea7026da2a91d09f325b5edbd
|
| 3 |
+
size 321729066
|
vector_db/metadata.pkl
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
-
size
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:6b859781f0a2bbefca6659c254d9fccbbca6e2b3d1b1c5790c83cd9ced3082bc
|
| 3 |
+
size 2701479
|