Commit
•
a50b299
1
Parent(s):
834b1c6
Fixing naming
Browse files- calc_params +0 -43
calc_params
DELETED
@@ -1,43 +0,0 @@
|
|
1 |
-
import math
|
2 |
-
|
3 |
-
# Helper function to pretty-print message sizes
|
4 |
-
def convert_params(params):
|
5 |
-
if params == 0:
|
6 |
-
return "0"
|
7 |
-
size_name = ("", "K", "M", "B", "T", "P", "E", "Z", "Y")
|
8 |
-
i = int(math.floor(math.log(params, 1000)))
|
9 |
-
p = math.pow(1000, i)
|
10 |
-
s = round(params / p, 2)
|
11 |
-
return "%s %s" % (s, size_name[i])
|
12 |
-
|
13 |
-
# Parameter Calculation function
|
14 |
-
def calc_params(vocab_size, tied_embeddings, hidden_size, sequence_length, num_layers, moe, num_experts, expert_interval, topk, ffn_expansion_factor, num_mlp_linears, kv_size_ratio):
|
15 |
-
if tied_embeddings:
|
16 |
-
embedding_params = hidden_size * vocab_size
|
17 |
-
else:
|
18 |
-
embedding_params = 2 * hidden_size * vocab_size
|
19 |
-
position_embedding_params = hidden_size * sequence_length
|
20 |
-
attention_params = int(2 * (1 + kv_size_ratio) * num_layers * hidden_size * hidden_size)
|
21 |
-
layernorm_params = 13 * num_layers * hidden_size
|
22 |
-
|
23 |
-
if moe:
|
24 |
-
num_expert_layers = num_layers / expert_interval
|
25 |
-
ffn_expert_params = num_mlp_linears * ffn_expansion_factor * num_expert_layers * num_experts * hidden_size * hidden_size
|
26 |
-
ffn_dense_params = num_mlp_linears * ffn_expansion_factor * (num_layers - num_expert_layers) * hidden_size * hidden_size
|
27 |
-
ffn_params = ffn_expert_params + ffn_dense_params
|
28 |
-
gating_params = num_expert_layers * hidden_size * num_experts
|
29 |
-
else:
|
30 |
-
ffn_params = num_mlp_linears * ffn_expansion_factor * num_layers * hidden_size * hidden_size
|
31 |
-
|
32 |
-
total_params = embedding_params + attention_params + ffn_params + position_embedding_params + layernorm_params
|
33 |
-
|
34 |
-
if moe:
|
35 |
-
total_params += gating_params
|
36 |
-
|
37 |
-
return f"""
|
38 |
-
Embedding parameters: {convert_params(embedding_params)}
|
39 |
-
Attention parameters: {convert_params(attention_params)}
|
40 |
-
FFN parameters: {convert_params(ffn_params)}
|
41 |
-
{'Gating parameters: ' + convert_params(gating_params) if moe else ''}
|
42 |
-
Total Params in the Model: {convert_params(total_params)}
|
43 |
-
"""
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|