kronos-trader/finetune_csv/README.md

Kronos Finetuning on Your Custom csv Dataset

Supports fine-tuning training with custom CSV data using configuration files

1. Quick Start

Configuration Setup

First edit the config.yaml file to set the correct paths and parameters:

# Data configuration
data:
  data_path: "/path/to/your/data.csv"
  lookback_window: 512
  predict_window: 48
  # ... other parameters

# Model path configuration
model_paths:
  pretrained_tokenizer: "/path/to/pretrained/tokenizer"
  pretrained_predictor: "/path/to/pretrained/predictor"
  base_save_path: "/path/to/save/models"
  # ... other paths

Run Training

Using train_sequential

# Complete training
python train_sequential.py --config configs/config_ali09988_candle-5min.yaml

# Skip existing models
python train_sequential.py --config configs/config_ali09988_candle-5min.yaml --skip-existing

# Only train tokenizer
python train_sequential.py --config configs/config_ali09988_candle-5min.yaml --skip-basemodel

# Only train basemodel
python train_sequential.py --config configs/config_ali09988_candle-5min.yaml --skip-tokenizer

Run each stage separately

# Only train tokenizer
python finetune_tokenizer.py --config configs/config_ali09988_candle-5min.yaml 

# Only train basemodel (requires fine-tuned tokenizer first)
python finetune_base_model.py --config configs/config_ali09988_candle-5min.yaml 

DDP Training

# Choose communication protocol yourself, nccl can be replaced with gloo
DIST_BACKEND=nccl \
torchrun --standalone --nproc_per_node=8 train_sequential.py --config configs/config_ali09988_candle-5min.yaml

2. Training Results

Data Format: Ensure CSV file contains the following columns: timestamps, open, high, low, close, volume, amount