jefsnacker
commited on
Commit
•
bfac513
1
Parent(s):
dcc1bc4
first name model
Browse files- first_name_gpt_config.yaml +45 -0
- first_name_gpt_weights.pt +3 -0
first_name_gpt_config.yaml
ADDED
@@ -0,0 +1,45 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
attn_pdrop: 0.1
|
2 |
+
b1: 0.9
|
3 |
+
b2: 0.95
|
4 |
+
batch_size: 4096
|
5 |
+
blocks: 6
|
6 |
+
d_model: 128
|
7 |
+
embd_pdrop: 0.1
|
8 |
+
epochs: 50000
|
9 |
+
heads: 4
|
10 |
+
lr: 0.0003
|
11 |
+
model_type: gpt
|
12 |
+
num_final_chars_in_dataset: 2
|
13 |
+
num_workers: 4
|
14 |
+
resid_pdrop: 0.1
|
15 |
+
stoi:
|
16 |
+
.: 0
|
17 |
+
a: 1
|
18 |
+
b: 2
|
19 |
+
c: 3
|
20 |
+
d: 4
|
21 |
+
e: 5
|
22 |
+
f: 6
|
23 |
+
g: 7
|
24 |
+
h: 8
|
25 |
+
i: 9
|
26 |
+
j: 10
|
27 |
+
k: 11
|
28 |
+
l: 12
|
29 |
+
m: 13
|
30 |
+
n: 14
|
31 |
+
o: 15
|
32 |
+
p: 16
|
33 |
+
q: 17
|
34 |
+
r: 18
|
35 |
+
s: 19
|
36 |
+
t: 20
|
37 |
+
u: 21
|
38 |
+
v: 22
|
39 |
+
w: 23
|
40 |
+
x: 24
|
41 |
+
y: 25
|
42 |
+
z: 26
|
43 |
+
vocab: 27
|
44 |
+
weight_decay: 0.1
|
45 |
+
window: 32
|
first_name_gpt_weights.pt
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:b165d7fe7bd9f130d68fbf059a76d1e404934ed692b328f1661e29ad5909b615
|
3 |
+
size 4857267
|