ProDa/requirements.txt at main · OpenRaiser/ProDa · GitHub

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
# ProDA core runtime requirements (Python 3.10)
#
# Scope of this file:
#   1. Run the FastAPI backend + React/Vite frontend integration.
#   2. Run ProDA's core data generation / diagnosis logic.
#   3. Keep the legacy Streamlit UI usable.
#
# Not included on purpose:
#   - LLaMA-Factory training stack: torch / transformers / datasets / accelerate /
#     peft / deepspeed / trl / etc.
#   - OpenCompass evaluation stack: opencompass / mmengine / modelscope / etc.
#
# Install those two external projects separately after this file, following their
# own installation instructions. ProDA calls them by configured local paths.

# -----------------------------
# FastAPI backend
# -----------------------------
fastapi==0.120.2
uvicorn==0.38.0
pydantic==2.12.5
python-multipart==0.0.20
psutil==7.0.0

# -----------------------------
# LLM API clients
# -----------------------------
openai==2.6.0
anthropic==0.71.0

# -----------------------------
# Document ingestion
# -----------------------------
pypdf==6.9.2
python-docx==1.2.0

# -----------------------------
# Data utilities used by ProDA / legacy UI
# -----------------------------
pandas==2.2.3

# -----------------------------
# Legacy Streamlit UI compatibility
# -----------------------------
streamlit==1.51.0

# -----------------------------
# Notes for optional local model chat
# -----------------------------
# The FineTuning model-chat window imports torch / transformers / peft lazily.
# They are expected to be available after installing LLaMA-Factory. If you do not
# install LLaMA-Factory in the same environment, the rest of ProDA still runs, but
# local checkpoint chat will report missing inference dependencies.