config.py

"""DataSense HF Space configuration — swap model IDs here after new checkpoints."""

import os
from pathlib import Path

ROOT = Path(__file__).resolve().parent
DATA_DIR = ROOT / "data"

# Base + LoRA adapter — SFT v1 (stable demo; see story.html § demo choice)
BASE_MODEL = os.environ.get("DATASENSE_BASE_MODEL", "unsloth/gemma-4-E2B-it")
ADAPTER_MODEL = os.environ.get(
    "DATASENSE_ADAPTER_MODEL",
    "sanjaymalladi/DataSense-Modal-E2B-SFT",
)

MAX_SEQ_LENGTH = int(os.environ.get("DATASENSE_MAX_SEQ_LENGTH", "2048"))
LOAD_IN_4BIT = os.environ.get("DATASENSE_LOAD_4BIT", "1") == "1"
AGENT_MAX_STEPS = int(os.environ.get("DATASENSE_MAX_STEPS", "8"))
AGENT_MAX_NEW_TOKENS = int(os.environ.get("DATASENSE_MAX_NEW_TOKENS", "768"))
AGENT_EXEC_TIMEOUT = int(os.environ.get("DATASENSE_EXEC_TIMEOUT", "30"))
AGENT_TEMPERATURE = float(os.environ.get("DATASENSE_TEMPERATURE", "0.2"))

DONE_MARKERS = ("**Summary:**", "**Finding:**", "**Conclusion:**", "**Results:**")