VQLM / gauss_mad_d1 /meta_000091.json
ASzecsenyi's picture
Upload gauss_mad_d1/meta_000091.json with huggingface_hub
ee04eb9 verified
{
"step": 91,
"val_bpb": 3.8118258814804027,
"model_config": {
"run": "gauss_mad_d1",
"device_type": "",
"depth": 1,
"aspect_ratio": 64,
"head_dim": 128,
"sequence_len": 256,
"vocab_size": 16,
"rva_blocks": [],
"gauss_blocks": [
0
],
"num_iterations": -1,
"target_flops": -1.0,
"target_param_data_ratio": 10,
"data_dir": "mad_data",
"device_batch_size": 32,
"total_batch_size": 8192,
"embedding_lr": 0.3,
"unembedding_lr": 0.004,
"grad_clip": 1.0,
"weight_decay": 0.0,
"matrix_lr": 0.02,
"adam_beta1": 0.8,
"adam_beta2": 0.95,
"warmup_ratio": 0.0,
"warmdown_ratio": 0.4,
"final_lr_frac": 0.0,
"resume_from_step": -1,
"eval_every": 250,
"eval_tokens": 65536,
"core_metric_every": -1,
"core_metric_max_per_task": 500,
"sample_every": -1,
"save_every": -1,
"push_checkpoints_to_hub": true,
"use_profiler": true,
"profile_step": 2,
"profile_micro_step": 0,
"memory_history_max_entries": 10000,
"model_tag": "gauss_mad_d1",
"n_layer": 1,
"n_head": 1,
"n_kv_head": 1,
"n_embd": 64,
"d_state": 32
},
"user_config": {
"run": "gauss_mad_d1",
"device_type": "",
"depth": 1,
"aspect_ratio": 64,
"head_dim": 128,
"sequence_len": 256,
"vocab_size": 16,
"rva_blocks": [],
"gauss_blocks": [
0
],
"num_iterations": -1,
"target_flops": -1.0,
"target_param_data_ratio": 10,
"data_dir": "mad_data",
"device_batch_size": 32,
"total_batch_size": 8192,
"embedding_lr": 0.3,
"unembedding_lr": 0.004,
"grad_clip": 1.0,
"weight_decay": 0.0,
"matrix_lr": 0.02,
"adam_beta1": 0.8,
"adam_beta2": 0.95,
"warmup_ratio": 0.0,
"warmdown_ratio": 0.4,
"final_lr_frac": 0.0,
"resume_from_step": -1,
"eval_every": 250,
"eval_tokens": 65536,
"core_metric_every": -1,
"core_metric_max_per_task": 500,
"sample_every": -1,
"save_every": -1,
"push_checkpoints_to_hub": true,
"use_profiler": true,
"profile_step": 2,
"profile_micro_step": 0,
"memory_history_max_entries": 10000,
"model_tag": "gauss_mad_d1"
},
"device_batch_size": 32,
"sequence_len": 256,
"dataloader_state_dict": {
"sample_idx": 2944
},
"loop_state": {
"min_val_bpb": 3.8118258814804027,
"smooth_train_loss": 2.642747494703387,
"total_training_time": 3.2108139991760254
}
}