|
{ |
|
"_name_or_path": "configs/mom_gated_deltanet_1.3B_4e2a1s.json", |
|
"architectures": [ |
|
"MomGatedDeltaNetForCausalLM" |
|
], |
|
"attn": null, |
|
"attn_mode": "chunk", |
|
"aux_loss_scale": 0.01, |
|
"bos_token_id": 1, |
|
"capacity": 1.0, |
|
"conv_size": 4, |
|
"eos_token_id": 2, |
|
"expand_v": 1, |
|
"fuse_cross_entropy": true, |
|
"head_dim": 256, |
|
"hidden_act": "swish", |
|
"hidden_ratio": 4, |
|
"hidden_size": 2048, |
|
"initializer_range": 0.02, |
|
"intermediate_size": null, |
|
"max_position_embeddings": 2048, |
|
"model_type": "mom_gated_deltanet", |
|
"norm_eps": 1e-06, |
|
"norm_first": false, |
|
"num_memories": 4, |
|
"num_heads": 4, |
|
"num_hidden_layers": 24, |
|
"shared_mem": true, |
|
"single_kv_proj": false, |
|
"tie_word_embeddings": true, |
|
"topk": 2, |
|
"torch_dtype": "bfloat16", |
|
"transformers_version": "4.47.1", |
|
"use_cache": true, |
|
"use_gate": true, |
|
"use_layer_wise_balance": true, |
|
"use_short_conv": true, |
|
"vocab_size": 32000 |
|
} |
|
|