- merge: upload `transformers` implementation (efd6eabee32eabd8675eae49359287ad061fe197) - feat: use latest modeling code (163f149452b3ecef0f4c3df5ce63c69652e45b78)
26 lines
609 B
JSON
26 lines
609 B
JSON
{
|
|
"architectures": [
|
|
"StableLmForCausalLM"
|
|
],
|
|
"bos_token_id": 100257,
|
|
"eos_token_id": 100257,
|
|
"hidden_act": "silu",
|
|
"hidden_size": 2048,
|
|
"initializer_range": 0.02,
|
|
"intermediate_size": 5632,
|
|
"max_position_embeddings": 4096,
|
|
"model_type": "stablelm",
|
|
"layer_norm_eps": 1e-05,
|
|
"num_attention_heads": 32,
|
|
"num_hidden_layers": 24,
|
|
"num_key_value_heads": 32,
|
|
"partial_rotary_factor": 0.25,
|
|
"rope_theta": 10000,
|
|
"tie_word_embeddings": false,
|
|
"torch_dtype": "bfloat16",
|
|
"transformers_version": "4.38.0",
|
|
"use_cache": true,
|
|
"use_qkv_bias": true,
|
|
"vocab_size": 100352
|
|
}
|