Rename model_config.yaml to config.json
Browse files- config.json +31 -0
- model_config.yaml +0 -28
config.json
ADDED
|
@@ -0,0 +1,31 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"bias": false,
|
| 3 |
+
"block_size": 2048,
|
| 4 |
+
"gelu_approximate": "none",
|
| 5 |
+
"head_size": 64,
|
| 6 |
+
"hf_config": {
|
| 7 |
+
"name": "TinyLlama-1.1B-intermediate-step-1431k-3T",
|
| 8 |
+
"org": "TinyLlama"
|
| 9 |
+
},
|
| 10 |
+
"intermediate_size": 5632,
|
| 11 |
+
"lm_head_bias": false,
|
| 12 |
+
"mlp_class_name": "LLaMAMLP",
|
| 13 |
+
"n_embd": 2048,
|
| 14 |
+
"n_expert": 0,
|
| 15 |
+
"n_expert_per_token": 0,
|
| 16 |
+
"n_head": 32,
|
| 17 |
+
"n_layer": 22,
|
| 18 |
+
"n_query_groups": 4,
|
| 19 |
+
"name": "tiny-llama-1.1b",
|
| 20 |
+
"norm_class_name": "RMSNorm",
|
| 21 |
+
"norm_eps": 1.0e-05,
|
| 22 |
+
"padded_vocab_size": 32000,
|
| 23 |
+
"padding_multiple": 64,
|
| 24 |
+
"parallel_residual": false,
|
| 25 |
+
"rope_base": 10000,
|
| 26 |
+
"rope_condense_ratio": 1,
|
| 27 |
+
"rotary_percentage": 1.0,
|
| 28 |
+
"scale_embeddings": false,
|
| 29 |
+
"shared_attention_norm": false,
|
| 30 |
+
"vocab_size": 32000
|
| 31 |
+
}
|
model_config.yaml
DELETED
|
@@ -1,28 +0,0 @@
|
|
| 1 |
-
bias: false
|
| 2 |
-
block_size: 2048
|
| 3 |
-
gelu_approximate: none
|
| 4 |
-
head_size: 64
|
| 5 |
-
hf_config:
|
| 6 |
-
name: TinyLlama-1.1B-intermediate-step-1431k-3T
|
| 7 |
-
org: TinyLlama
|
| 8 |
-
intermediate_size: 5632
|
| 9 |
-
lm_head_bias: false
|
| 10 |
-
mlp_class_name: LLaMAMLP
|
| 11 |
-
n_embd: 2048
|
| 12 |
-
n_expert: 0
|
| 13 |
-
n_expert_per_token: 0
|
| 14 |
-
n_head: 32
|
| 15 |
-
n_layer: 22
|
| 16 |
-
n_query_groups: 4
|
| 17 |
-
name: tiny-llama-1.1b
|
| 18 |
-
norm_class_name: RMSNorm
|
| 19 |
-
norm_eps: 1.0e-05
|
| 20 |
-
padded_vocab_size: 32000
|
| 21 |
-
padding_multiple: 64
|
| 22 |
-
parallel_residual: false
|
| 23 |
-
rope_base: 10000
|
| 24 |
-
rope_condense_ratio: 1
|
| 25 |
-
rotary_percentage: 1.0
|
| 26 |
-
scale_embeddings: false
|
| 27 |
-
shared_attention_norm: false
|
| 28 |
-
vocab_size: 32000
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|