Upload config.json with huggingface_hub
Browse files- config.json +12 -5
config.json
CHANGED
|
@@ -1,7 +1,7 @@
|
|
| 1 |
{
|
| 2 |
"activation_function": "gelu_new",
|
| 3 |
"architectures": [
|
| 4 |
-
"
|
| 5 |
],
|
| 6 |
"attn_pdrop": 0.1,
|
| 7 |
"bos_token_id": 50256,
|
|
@@ -9,11 +9,11 @@
|
|
| 9 |
"eos_token_id": 50256,
|
| 10 |
"initializer_range": 0.02,
|
| 11 |
"layer_norm_epsilon": 1e-05,
|
| 12 |
-
"model_type": "
|
| 13 |
"n_ctx": 1024,
|
| 14 |
"n_embd": 768,
|
| 15 |
"n_head": 12,
|
| 16 |
-
"n_inner":
|
| 17 |
"n_layer": 12,
|
| 18 |
"n_positions": 1024,
|
| 19 |
"reorder_and_upcast_attn": false,
|
|
@@ -34,5 +34,12 @@
|
|
| 34 |
"torch_dtype": "float32",
|
| 35 |
"transformers_version": "4.55.2",
|
| 36 |
"use_cache": true,
|
| 37 |
-
"vocab_size": 50257
|
| 38 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
{
|
| 2 |
"activation_function": "gelu_new",
|
| 3 |
"architectures": [
|
| 4 |
+
"RecursiveGPT2LMHeadModel"
|
| 5 |
],
|
| 6 |
"attn_pdrop": 0.1,
|
| 7 |
"bos_token_id": 50256,
|
|
|
|
| 9 |
"eos_token_id": 50256,
|
| 10 |
"initializer_range": 0.02,
|
| 11 |
"layer_norm_epsilon": 1e-05,
|
| 12 |
+
"model_type": "recursive_gpt2",
|
| 13 |
"n_ctx": 1024,
|
| 14 |
"n_embd": 768,
|
| 15 |
"n_head": 12,
|
| 16 |
+
"n_inner": 3072,
|
| 17 |
"n_layer": 12,
|
| 18 |
"n_positions": 1024,
|
| 19 |
"reorder_and_upcast_attn": false,
|
|
|
|
| 34 |
"torch_dtype": "float32",
|
| 35 |
"transformers_version": "4.55.2",
|
| 36 |
"use_cache": true,
|
| 37 |
+
"vocab_size": 50257,
|
| 38 |
+
"K": 2,
|
| 39 |
+
"rank": 8,
|
| 40 |
+
"auto_map": {
|
| 41 |
+
"AutoModelForCausalLM": "RecursiveGPT2Model.RecursiveGPT2LMHeadModel",
|
| 42 |
+
"AutoConfig": "RecursiveGPT2Model.RecursiveGPT2Config"
|
| 43 |
+
},
|
| 44 |
+
"trust_remote_code": true
|
| 45 |
+
}
|