limloop/whiff-mamba2-20M
189
1{2 "architectures": [3 "Mamba2ForCausalLM"4 ],5 "bos_token_id": 6,6 "chunk_size": 256,7 "conv_kernel": 4,8 "eos_token_id": 6,9 "expand": 1.5,10 "head_dim": 64,11 "hidden_act": "silu",12 "hidden_size": 512,13 "initializer_range": 0.1,14 "layer_norm_epsilon": 1e-05,15 "model_type": "mamba2",16 "n_groups": 2,17 "num_heads": 12,18 "num_hidden_layers": 9,19 "pad_token_id": 7,20 "rescale_prenorm_residual": false,21 "residual_in_fp32": true,22 "rms_norm": true,23 "state_size": 64,24 "tie_word_embeddings": false,25 "time_step_floor": 0.0001,26 "time_step_limit": [27 0.0,28 {29 "__float__": "Infinity"30 }31 ],32 "time_step_max": 0.1,33 "time_step_min": 0.001,34 "time_step_rank": 32,35 "torch_dtype": "float32",36 "transformers_version": "4.54.1",37 "use_bias": false,38 "use_cache": true,39 "use_conv_bias": true,40 "vocab_size": 819241}42 