diffusers/FLUX.1-dev-bnb-4bit
62.4k
1{2 "architectures": [3 "T5EncoderModel"4 ],5 "classifier_dropout": 0.0,6 "d_ff": 10240,7 "d_kv": 64,8 "d_model": 4096,9 "decoder_start_token_id": 0,10 "dense_act_fn": "gelu_new",11 "dropout_rate": 0.1,12 "eos_token_id": 1,13 "feed_forward_proj": "gated-gelu",14 "initializer_factor": 1.0,15 "is_encoder_decoder": true,16 "is_gated_act": true,17 "layer_norm_epsilon": 1e-06,18 "model_type": "t5",19 "num_decoder_layers": 24,20 "num_heads": 64,21 "num_layers": 24,22 "output_past": true,23 "pad_token_id": 0,24 "quantization_config": {25 "_load_in_4bit": true,26 "_load_in_8bit": false,27 "bnb_4bit_compute_dtype": "bfloat16",28 "bnb_4bit_quant_storage": "uint8",29 "bnb_4bit_quant_type": "nf4",30 "bnb_4bit_use_double_quant": false,31 "llm_int8_enable_fp32_cpu_offload": false,32 "llm_int8_has_fp16_weight": false,33 "llm_int8_skip_modules": null,34 "llm_int8_threshold": 6.0,35 "load_in_4bit": true,36 "load_in_8bit": false,37 "quant_method": "bitsandbytes"38 },39 "relative_attention_max_distance": 128,40 "relative_attention_num_buckets": 32,41 "tie_word_embeddings": false,42 "torch_dtype": "bfloat16",43 "transformers_version": "4.52.0.dev0",44 "use_cache": true,45 "vocab_size": 3212846}47 