| { | |
| "bits": 4, | |
| "group_size": 128, | |
| "sym": false, | |
| "data_type": "int", | |
| "enable_quanted_input": true, | |
| "enable_minmax_tuning": true, | |
| "seqlen": 2048, | |
| "train_bs": 8, | |
| "scale_dtype": "torch.float16", | |
| "lr": 0.001, | |
| "minmax_lr": 0.001, | |
| "gradient_accumulate_steps": 1, | |
| "iters": 1000, | |
| "amp": true, | |
| "nsamples": 512, | |
| "low_gpu_mem_usage": false, | |
| "dataset": "NeelNanda/pile-10k", | |
| "autoround_version": "0.3.0.dev", | |
| "quant_method": "intel/auto-round", | |
| "backend": "auto_round:exllamav2", | |
| "extra_config": { | |
| "model.layers.0.mlp.shared_expert_gate": { | |
| "data_type": "bfloat", | |
| "bits": 32, | |
| "group_size": null, | |
| "sym": null | |
| }, | |
| "model.layers.1.mlp.shared_expert_gate": { | |
| "data_type": "bfloat", | |
| "bits": 32, | |
| "group_size": null, | |
| "sym": null | |
| }, | |
| "model.layers.2.mlp.shared_expert_gate": { | |
| "data_type": "bfloat", | |
| "bits": 32, | |
| "group_size": null, | |
| "sym": null | |
| }, | |
| "model.layers.3.mlp.shared_expert_gate": { | |
| "data_type": "bfloat", | |
| "bits": 32, | |
| "group_size": null, | |
| "sym": null | |
| }, | |
| "model.layers.4.mlp.shared_expert_gate": { | |
| "data_type": "bfloat", | |
| "bits": 32, | |
| "group_size": null, | |
| "sym": null | |
| }, | |
| "model.layers.5.mlp.shared_expert_gate": { | |
| "data_type": "bfloat", | |
| "bits": 32, | |
| "group_size": null, | |
| "sym": null | |
| }, | |
| "model.layers.6.mlp.shared_expert_gate": { | |
| "data_type": "bfloat", | |
| "bits": 32, | |
| "group_size": null, | |
| "sym": null | |
| }, | |
| "model.layers.7.mlp.shared_expert_gate": { | |
| "data_type": "bfloat", | |
| "bits": 32, | |
| "group_size": null, | |
| "sym": null | |
| }, | |
| "model.layers.8.mlp.shared_expert_gate": { | |
| "data_type": "bfloat", | |
| "bits": 32, | |
| "group_size": null, | |
| "sym": null | |
| }, | |
| "model.layers.9.mlp.shared_expert_gate": { | |
| "data_type": "bfloat", | |
| "bits": 32, | |
| "group_size": null, | |
| "sym": null | |
| }, | |
| "model.layers.10.mlp.shared_expert_gate": { | |
| "data_type": "bfloat", | |
| "bits": 32, | |
| "group_size": null, | |
| "sym": null | |
| }, | |
| "model.layers.11.mlp.shared_expert_gate": { | |
| "data_type": "bfloat", | |
| "bits": 32, | |
| "group_size": null, | |
| "sym": null | |
| }, | |
| "model.layers.12.mlp.shared_expert_gate": { | |
| "data_type": "bfloat", | |
| "bits": 32, | |
| "group_size": null, | |
| "sym": null | |
| }, | |
| "model.layers.13.mlp.shared_expert_gate": { | |
| "data_type": "bfloat", | |
| "bits": 32, | |
| "group_size": null, | |
| "sym": null | |
| }, | |
| "model.layers.14.mlp.shared_expert_gate": { | |
| "data_type": "bfloat", | |
| "bits": 32, | |
| "group_size": null, | |
| "sym": null | |
| }, | |
| "model.layers.15.mlp.shared_expert_gate": { | |
| "data_type": "bfloat", | |
| "bits": 32, | |
| "group_size": null, | |
| "sym": null | |
| }, | |
| "model.layers.16.mlp.shared_expert_gate": { | |
| "data_type": "bfloat", | |
| "bits": 32, | |
| "group_size": null, | |
| "sym": null | |
| }, | |
| "model.layers.17.mlp.shared_expert_gate": { | |
| "data_type": "bfloat", | |
| "bits": 32, | |
| "group_size": null, | |
| "sym": null | |
| }, | |
| "model.layers.18.mlp.shared_expert_gate": { | |
| "data_type": "bfloat", | |
| "bits": 32, | |
| "group_size": null, | |
| "sym": null | |
| }, | |
| "model.layers.19.mlp.shared_expert_gate": { | |
| "data_type": "bfloat", | |
| "bits": 32, | |
| "group_size": null, | |
| "sym": null | |
| }, | |
| "model.layers.20.mlp.shared_expert_gate": { | |
| "data_type": "bfloat", | |
| "bits": 32, | |
| "group_size": null, | |
| "sym": null | |
| }, | |
| "model.layers.21.mlp.shared_expert_gate": { | |
| "data_type": "bfloat", | |
| "bits": 32, | |
| "group_size": null, | |
| "sym": null | |
| }, | |
| "model.layers.22.mlp.shared_expert_gate": { | |
| "data_type": "bfloat", | |
| "bits": 32, | |
| "group_size": null, | |
| "sym": null | |
| }, | |
| "model.layers.23.mlp.shared_expert_gate": { | |
| "data_type": "bfloat", | |
| "bits": 32, | |
| "group_size": null, | |
| "sym": null | |
| }, | |
| "model.layers.24.mlp.shared_expert_gate": { | |
| "data_type": "bfloat", | |
| "bits": 32, | |
| "group_size": null, | |
| "sym": null | |
| }, | |
| "model.layers.25.mlp.shared_expert_gate": { | |
| "data_type": "bfloat", | |
| "bits": 32, | |
| "group_size": null, | |
| "sym": null | |
| }, | |
| "model.layers.26.mlp.shared_expert_gate": { | |
| "data_type": "bfloat", | |
| "bits": 32, | |
| "group_size": null, | |
| "sym": null | |
| }, | |
| "model.layers.27.mlp.shared_expert_gate": { | |
| "data_type": "bfloat", | |
| "bits": 32, | |
| "group_size": null, | |
| "sym": null | |
| }, | |
| "model.layers.0.mlp.gate": { | |
| "data_type": "bfloat", | |
| "bits": 32, | |
| "group_size": null, | |
| "sym": null | |
| }, | |
| "model.layers.1.mlp.gate": { | |
| "data_type": "bfloat", | |
| "bits": 32, | |
| "group_size": null, | |
| "sym": null | |
| }, | |
| "model.layers.2.mlp.gate": { | |
| "data_type": "bfloat", | |
| "bits": 32, | |
| "group_size": null, | |
| "sym": null | |
| }, | |
| "model.layers.3.mlp.gate": { | |
| "data_type": "bfloat", | |
| "bits": 32, | |
| "group_size": null, | |
| "sym": null | |
| }, | |
| "model.layers.4.mlp.gate": { | |
| "data_type": "bfloat", | |
| "bits": 32, | |
| "group_size": null, | |
| "sym": null | |
| }, | |
| "model.layers.5.mlp.gate": { | |
| "data_type": "bfloat", | |
| "bits": 32, | |
| "group_size": null, | |
| "sym": null | |
| }, | |
| "model.layers.6.mlp.gate": { | |
| "data_type": "bfloat", | |
| "bits": 32, | |
| "group_size": null, | |
| "sym": null | |
| }, | |
| "model.layers.7.mlp.gate": { | |
| "data_type": "bfloat", | |
| "bits": 32, | |
| "group_size": null, | |
| "sym": null | |
| }, | |
| "model.layers.8.mlp.gate": { | |
| "data_type": "bfloat", | |
| "bits": 32, | |
| "group_size": null, | |
| "sym": null | |
| }, | |
| "model.layers.9.mlp.gate": { | |
| "data_type": "bfloat", | |
| "bits": 32, | |
| "group_size": null, | |
| "sym": null | |
| }, | |
| "model.layers.10.mlp.gate": { | |
| "data_type": "bfloat", | |
| "bits": 32, | |
| "group_size": null, | |
| "sym": null | |
| }, | |
| "model.layers.11.mlp.gate": { | |
| "data_type": "bfloat", | |
| "bits": 32, | |
| "group_size": null, | |
| "sym": null | |
| }, | |
| "model.layers.12.mlp.gate": { | |
| "data_type": "bfloat", | |
| "bits": 32, | |
| "group_size": null, | |
| "sym": null | |
| }, | |
| "model.layers.13.mlp.gate": { | |
| "data_type": "bfloat", | |
| "bits": 32, | |
| "group_size": null, | |
| "sym": null | |
| }, | |
| "model.layers.14.mlp.gate": { | |
| "data_type": "bfloat", | |
| "bits": 32, | |
| "group_size": null, | |
| "sym": null | |
| }, | |
| "model.layers.15.mlp.gate": { | |
| "data_type": "bfloat", | |
| "bits": 32, | |
| "group_size": null, | |
| "sym": null | |
| }, | |
| "model.layers.16.mlp.gate": { | |
| "data_type": "bfloat", | |
| "bits": 32, | |
| "group_size": null, | |
| "sym": null | |
| }, | |
| "model.layers.17.mlp.gate": { | |
| "data_type": "bfloat", | |
| "bits": 32, | |
| "group_size": null, | |
| "sym": null | |
| }, | |
| "model.layers.18.mlp.gate": { | |
| "data_type": "bfloat", | |
| "bits": 32, | |
| "group_size": null, | |
| "sym": null | |
| }, | |
| "model.layers.19.mlp.gate": { | |
| "data_type": "bfloat", | |
| "bits": 32, | |
| "group_size": null, | |
| "sym": null | |
| }, | |
| "model.layers.20.mlp.gate": { | |
| "data_type": "bfloat", | |
| "bits": 32, | |
| "group_size": null, | |
| "sym": null | |
| }, | |
| "model.layers.21.mlp.gate": { | |
| "data_type": "bfloat", | |
| "bits": 32, | |
| "group_size": null, | |
| "sym": null | |
| }, | |
| "model.layers.22.mlp.gate": { | |
| "data_type": "bfloat", | |
| "bits": 32, | |
| "group_size": null, | |
| "sym": null | |
| }, | |
| "model.layers.23.mlp.gate": { | |
| "data_type": "bfloat", | |
| "bits": 32, | |
| "group_size": null, | |
| "sym": null | |
| }, | |
| "model.layers.24.mlp.gate": { | |
| "data_type": "bfloat", | |
| "bits": 32, | |
| "group_size": null, | |
| "sym": null | |
| }, | |
| "model.layers.25.mlp.gate": { | |
| "data_type": "bfloat", | |
| "bits": 32, | |
| "group_size": null, | |
| "sym": null | |
| }, | |
| "model.layers.26.mlp.gate": { | |
| "data_type": "bfloat", | |
| "bits": 32, | |
| "group_size": null, | |
| "sym": null | |
| }, | |
| "model.layers.27.mlp.gate": { | |
| "data_type": "bfloat", | |
| "bits": 32, | |
| "group_size": null, | |
| "sym": null | |
| } | |
| } | |
| } |