tinyLM-8M-exp / config.json
User01110's picture
Use custom config type to prevent Qwen3 fallback
9632173 verified
Raw
History Blame Contribute Delete
844 Bytes
{
"model_type": "tinyqwen3_novelty",
"architectures": [
"TinyQwen3NoveltyForCausalLM"
],
"auto_map": {
"AutoConfig": "modeling_tinyqwen3_novelty.TinyQwen3NoveltyConfig",
"AutoModelForCausalLM": "modeling_tinyqwen3_novelty.TinyQwen3NoveltyForCausalLM"
},
"vocab_size": 4098,
"hidden_size": 256,
"intermediate_size": 896,
"num_hidden_layers": 8,
"num_attention_heads": 8,
"num_key_value_heads": 4,
"head_dim": 32,
"rms_norm_eps": 1e-06,
"rope_theta": 2500.0,
"max_position_embeddings": 512,
"tie_word_embeddings": true,
"initializer_range": 0.02,
"torch_dtype": "float32",
"bos_token_id": 1,
"eos_token_id": 2,
"pad_token_id": 2,
"novelty_gate_floor": 0.05,
"novelty_gate_type": "math_rms_abs_delta",
"im_start_token_id": 4096,
"im_end_token_id": 4097
}