Hezar: Upload model_config.yaml
Browse files- model_config.yaml +33 -0
model_config.yaml
ADDED
@@ -0,0 +1,33 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
name: gpt2_text_generation
|
2 |
+
config_type: model
|
3 |
+
add_cross_attention: false
|
4 |
+
vocab_size: 42001
|
5 |
+
attn_pdrop: 0.1
|
6 |
+
bos_token_id: 5
|
7 |
+
embd_pdrop: 0.1
|
8 |
+
eos_token_id: 5
|
9 |
+
gradient_checkpointing: false
|
10 |
+
initializer_range: 0.02
|
11 |
+
layer_norm_epsilon: 1.0e-05
|
12 |
+
model_type: gpt2
|
13 |
+
n_ctx: 1024
|
14 |
+
n_embd: 768
|
15 |
+
n_head: 12
|
16 |
+
n_layer: 12
|
17 |
+
n_positions: 1024
|
18 |
+
resid_pdrop: 0.1
|
19 |
+
summary_activation: false
|
20 |
+
summary_first_dropout: 0.1
|
21 |
+
use_cache: true
|
22 |
+
generation:
|
23 |
+
config_type: model
|
24 |
+
bos_token_id: 0
|
25 |
+
do_sample: true
|
26 |
+
decoder_start_token_id: 0
|
27 |
+
early_stopping: true
|
28 |
+
eos_token_id: 2
|
29 |
+
length_penalty: 2.0
|
30 |
+
max_new_tokens: 50
|
31 |
+
no_repeat_ngram_size: 3
|
32 |
+
num_beams: 4
|
33 |
+
pad_token_id: 1
|