Upload config.json with huggingface_hub
Browse files- config.json +45 -0
config.json
ADDED
|
@@ -0,0 +1,45 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"activation_type": "swiglu",
|
| 3 |
+
"alibi": false,
|
| 4 |
+
"alibi_bias_max": 8.0,
|
| 5 |
+
"architectures": [
|
| 6 |
+
"OLMoModelForCausalLM"
|
| 7 |
+
],
|
| 8 |
+
"attention_dropout": 0.0,
|
| 9 |
+
"attention_layer_norm": false,
|
| 10 |
+
"attention_layer_norm_with_affine": false,
|
| 11 |
+
"bias_for_layer_norm": false,
|
| 12 |
+
"block_group_size": 1,
|
| 13 |
+
"block_type": "sequential",
|
| 14 |
+
"clip_qkv": null,
|
| 15 |
+
"d_model": 2048,
|
| 16 |
+
"embedding_dropout": 0.0,
|
| 17 |
+
"embedding_size": 50304,
|
| 18 |
+
"eos_token_id": 50279,
|
| 19 |
+
"flash_attention": true,
|
| 20 |
+
"include_bias": false,
|
| 21 |
+
"init_cutoff_factor": null,
|
| 22 |
+
"init_device": "meta",
|
| 23 |
+
"init_fn": "mitchell",
|
| 24 |
+
"init_std": 0.02,
|
| 25 |
+
"layer_norm_type": "default",
|
| 26 |
+
"layer_norm_with_affine": false,
|
| 27 |
+
"max_sequence_length": 2048,
|
| 28 |
+
"mlp_hidden_size": null,
|
| 29 |
+
"mlp_ratio": 8,
|
| 30 |
+
"model_type": "olmo",
|
| 31 |
+
"multi_query_attention": false,
|
| 32 |
+
"n_heads": 16,
|
| 33 |
+
"n_kv_heads": null,
|
| 34 |
+
"n_layers": 16,
|
| 35 |
+
"pad_token_id": 1,
|
| 36 |
+
"precision": "amp_bf16",
|
| 37 |
+
"residual_dropout": 0.0,
|
| 38 |
+
"rope": true,
|
| 39 |
+
"rope_full_precision": true,
|
| 40 |
+
"scale_logits": false,
|
| 41 |
+
"transformers_version": "4.39.3",
|
| 42 |
+
"use_cache": true,
|
| 43 |
+
"vocab_size": 50280,
|
| 44 |
+
"weight_tying": true
|
| 45 |
+
}
|