Upload config_cb.json with huggingface_hub
Browse files- config_cb.json +31 -0
config_cb.json
ADDED
|
@@ -0,0 +1,31 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"vocab_size": 84992,
|
| 3 |
+
"embd_pdrop": 0.0,
|
| 4 |
+
"tie_word_embeddings": true,
|
| 5 |
+
"position_embedding_type": "alibi",
|
| 6 |
+
"n_positions": 2048,
|
| 7 |
+
"rotary_dim": null,
|
| 8 |
+
"alibi_scaling": null,
|
| 9 |
+
"n_embd": 1088,
|
| 10 |
+
"n_layer": 14,
|
| 11 |
+
"resid_pdrop": 0.0,
|
| 12 |
+
"layer_norm_epsilon": 1e-05,
|
| 13 |
+
"n_head": 17,
|
| 14 |
+
"scale_attn_weights": true,
|
| 15 |
+
"attn_pdrop": 0.0,
|
| 16 |
+
"scale_attn_by_inverse_layer_idx": false,
|
| 17 |
+
"n_inner": 2912,
|
| 18 |
+
"activation_function": "swiglu",
|
| 19 |
+
"mup_embeddings_scale": 9.1705785388303,
|
| 20 |
+
"mup_scale_qk_dot_by_d": true,
|
| 21 |
+
"mup_output_alpha": 1.09518349815769,
|
| 22 |
+
"mup_width_scale": 0.23529411764705882,
|
| 23 |
+
"initializer_range": 0.02,
|
| 24 |
+
"model_type": "btlm",
|
| 25 |
+
"use_cache": true,
|
| 26 |
+
"auto_map": {
|
| 27 |
+
"AutoConfig": "cerebras/btlm-3b-8k-base--configuration_btlm.BTLMConfig",
|
| 28 |
+
"AutoModel": "cerebras/btlm-3b-8k-base--modeling_btlm.BTLMModel",
|
| 29 |
+
"AutoModelForCausalLM": "cerebras/btlm-3b-8k-base--modeling_btlm.BTLMLMHeadModel"
|
| 30 |
+
}
|
| 31 |
+
}
|