mixer-multitask-2024 / config.json
serkanog's picture
Upload 5 files
82b8bb6 verified
Raw
History Blame Contribute Delete
433 Bytes
{
"architectures": [
"CustomResearchModel"
],
"architecture": "mixer",
"model_type": "mixer",
"hidden_size": 256,
"num_hidden_layers": 6,
"num_attention_heads": 2,
"intermediate_size": 512,
"hidden_act": "gelu_tanh",
"max_position_embeddings": 256,
"layer_norm_eps": 1e-12,
"checkpoint_status": "initialization-only",
"notes": "Untrained checkpoint for smoke tests; no benchmark claim."
}