sansar-20m / config.json
ks2812's picture
v0.1.0: tokv3_v01_d20m (ex-Gita 0.7177)
329a89b verified
Raw History Blame Contribute Delete
780 Bytes
{
"architectures": [
"SansarForCausalLM"
],
"auto_map": {
"AutoConfig": "configuration_sansar.SansarConfig",
"AutoModelForCausalLM": "modeling_sansar.SansarForCausalLM"
},
"bias": false,
"block_size": 512,
"bos_token_id": 1,
"dtype": "bfloat16",
"eos_token_id": 2,
"initializer_range": 0.02,
"layer_norm_eps": 1e-05,
"logit_cap": 0.0,
"mlp_ratio": 4,
"model_type": "sansar",
"n_embd": 512,
"n_head": 8,
"n_layer": 6,
"pad_token_id": 2,
"qk_norm": true,
"relu_sq": true,
"rope": true,
"rope_theta": 10000.0,
"tie_word_embeddings": false,
"torch_dtype": "bfloat16",
"transformers_version": "5.18.0",
"untied_head": true,
"use_cache": false,
"value_embed": false,
"value_embed_layers": "",
"vocab_size": 8000
}