Upload train_qwen_sovereign.py with huggingface_hub
Browse files- train_qwen_sovereign.py +0 -3
train_qwen_sovereign.py
CHANGED
|
@@ -1,7 +1,6 @@
|
|
| 1 |
#!/usr/bin/env python3
|
| 2 |
import os
|
| 3 |
import torch
|
| 4 |
-
import trackio
|
| 5 |
from datasets import load_dataset
|
| 6 |
from peft import LoraConfig
|
| 7 |
from trl import SFTTrainer, SFTConfig
|
|
@@ -42,7 +41,6 @@ def get_peft_config():
|
|
| 42 |
def train_model(model_id, run_name, hub_model_id):
|
| 43 |
"""Executes the training job."""
|
| 44 |
print(f"📦 Starting training for {model_id}...")
|
| 45 |
-
trackio.init(experiment="sovereign-qwen-finetuning")
|
| 46 |
|
| 47 |
# Configure 4-bit loading for memory efficiency on A10G/T4
|
| 48 |
bnb_config = BitsAndBytesConfig(
|
|
@@ -86,7 +84,6 @@ def train_model(model_id, run_name, hub_model_id):
|
|
| 86 |
print("🚀 Starting training...")
|
| 87 |
trainer.train()
|
| 88 |
trainer.push_to_hub()
|
| 89 |
-
trackio.finish()
|
| 90 |
print("✅ Training complete.")
|
| 91 |
|
| 92 |
if __name__ == "__main__":
|
|
|
|
| 1 |
#!/usr/bin/env python3
|
| 2 |
import os
|
| 3 |
import torch
|
|
|
|
| 4 |
from datasets import load_dataset
|
| 5 |
from peft import LoraConfig
|
| 6 |
from trl import SFTTrainer, SFTConfig
|
|
|
|
| 41 |
def train_model(model_id, run_name, hub_model_id):
|
| 42 |
"""Executes the training job."""
|
| 43 |
print(f"📦 Starting training for {model_id}...")
|
|
|
|
| 44 |
|
| 45 |
# Configure 4-bit loading for memory efficiency on A10G/T4
|
| 46 |
bnb_config = BitsAndBytesConfig(
|
|
|
|
| 84 |
print("🚀 Starting training...")
|
| 85 |
trainer.train()
|
| 86 |
trainer.push_to_hub()
|
|
|
|
| 87 |
print("✅ Training complete.")
|
| 88 |
|
| 89 |
if __name__ == "__main__":
|