Upload train2.py with huggingface_hub

train2.py

@@ -0,0 +1,35 @@
+from datasets import load_dataset
+from transformers import AutoModelForCausalLM, AutoTokenizer, TrainingArguments, BitsAndBytesConfig
+from peft import LoraConfig, get_peft_model, prepare_model_for_kbit_training
+from trl import SFTTrainer
+from huggingface_hub import login
+import torch
+import os
+
+# Login first
+login(token=os.environ.get("HF_TOKEN", ""))
+
+dataset = load_dataset("Achilles1089/achilles-web3-training")
+print("Loaded", len(dataset["train"]), "examples")
+
+bnb = BitsAndBytesConfig(load_in_4bit=True, bnb_4bit_quant_type="nf4", bnb_4bit_compute_dtype=torch.bfloat16)
+tok = AutoTokenizer.from_pretrained("Qwen/Qwen3-30B-A3B-Instruct-2507", trust_remote_code=True)
+tok.pad_token = tok.eos_token
+
+model = AutoModelForCausalLM.from_pretrained("Qwen/Qwen3-30B-A3B-Instruct-2507", quantization_config=bnb, device_map="auto", trust_remote_code=True)
+model = prepare_model_for_kbit_training(model)
+model = get_peft_model(model, LoraConfig(r=64, lora_alpha=128, target_modules=["q_proj","k_proj","v_proj","o_proj","gate_proj","up_proj","down_proj"], lora_dropout=0.05, task_type="CAUSAL_LM"))
+
+def fmt(ex):
+    t=""
+    for m in ex.get("messages",[]):
+        r,c=m.get("role",""),m.get("content","")
+        if r=="system": t+="<|im_start|>system\n"+c+"<|im_end|>\n"
+        elif r=="user": t+="<|im_start|>user\n"+c+"<|im_end|>\n"
+        elif r=="assistant": t+="<|im_start|>assistant\n"+c+"<|im_end|>\n"
+    return {"text":t}
+
+trainer = SFTTrainer(model=model, args=TrainingArguments(output_dir="./achilles", num_train_epochs=2, per_device_train_batch_size=2, gradient_accumulation_steps=4, learning_rate=2e-4, bf16=True, gradient_checkpointing=True, logging_steps=10, save_steps=500, report_to="none"), train_dataset=dataset["train"].map(fmt), eval_dataset=dataset["valid"].map(fmt), tokenizer=tok, dataset_text_field="text", max_seq_length=2048, packing=True)
+trainer.train()
+trainer.push_to_hub("Achilles1089/achilles-web3-30b")
+print("DONE!")