args-lambda-quant.json

{"quantization": "GPTQ-Int8", "model": "meta-llama/Llama-3.3-70B-Instruct", "dataset": "HuggingFaceH4/ultrachat_200k", "dataset_split": "train_sft", "dataset_name": null, "num_samples": 512, "seq_length": 2048, "batch_size": 32, "update_metadata_only": false}