Note: Please do not remove the questions. Answer beside them.
- Did you update?
pip install --upgrade unsloth unsloth_zoo Ans. Yes
Colab or Kaggle or local / cloud Ans. Local- Number GPUs used, use
nvidia-smi Ans. 1x RTX 5050 Laptop GPU, 8GB VRAM
- Which notebook? Please link! Ans. Custom training script (no notebook)
- Which Unsloth version, TRL version, transformers version, PyTorch version?
Ans. unsloth==2026.4.4, unsloth_zoo==2026.4.3, trl==0.23.0, transformers==5.5.0, torch==2.10.0, CUDA 12.8
- Which trainer?
SFTTrainer, GRPOTrainer etc Ans. GRPOTrainer with PatchFastRL("GRPO", FastLanguageModel)
from unsloth import FastLanguageModel, PatchFastRL
PatchFastRL("GRPO", FastLanguageModel)
from trl import GRPOConfig, GRPOTrainer
model, tokenizer = FastLanguageModel.from_pretrained(
model_name="unsloth/Llama-3.2-3B-Instruct-bnb-4bit",
max_seq_length=1024,
dtype=None,
load_in_4bit=True,
fast_inference=False,
)
model = FastLanguageModel.get_peft_model(
model, r=16,
target_modules=["q_proj","k_proj","v_proj","o_proj","gate_proj","up_proj","down_proj"],
lora_alpha=16, lora_dropout=0, bias="none",
use_gradient_checkpointing="unsloth",
)
Note: Please do not remove the questions. Answer beside them.
pip install --upgrade unsloth unsloth_zooAns. YesColaborKaggleor local / cloud Ans. Localnvidia-smiAns. 1x RTX 5050 Laptop GPU, 8GB VRAMAns. unsloth==2026.4.4, unsloth_zoo==2026.4.3, trl==0.23.0, transformers==5.5.0, torch==2.10.0, CUDA 12.8
SFTTrainer,GRPOTraineretc Ans. GRPOTrainer with PatchFastRL("GRPO", FastLanguageModel)