Upload model
Browse files- config.json +0 -15
- modeling_gpt.py +1 -2
config.json
CHANGED
@@ -7,21 +7,6 @@
|
|
7 |
"AutoModelForCausalLM": "modeling_gpt.GPTModelForTextGeneration"
|
8 |
},
|
9 |
"block_size": 1024,
|
10 |
-
"custom_pipelines": {
|
11 |
-
"text-generation": {
|
12 |
-
"default": {
|
13 |
-
"model": {
|
14 |
-
"pt": "samkeet/GPT_124M"
|
15 |
-
}
|
16 |
-
},
|
17 |
-
"impl": "pipeline_gpt.GPT124MTextGenerationPipeline",
|
18 |
-
"pt": [
|
19 |
-
"AutoModelForCausalLM"
|
20 |
-
],
|
21 |
-
"tf": [],
|
22 |
-
"type": "text"
|
23 |
-
}
|
24 |
-
},
|
25 |
"model_type": "custom_gpt",
|
26 |
"n_embd": 768,
|
27 |
"n_head": 12,
|
|
|
7 |
"AutoModelForCausalLM": "modeling_gpt.GPTModelForTextGeneration"
|
8 |
},
|
9 |
"block_size": 1024,
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
10 |
"model_type": "custom_gpt",
|
11 |
"n_embd": 768,
|
12 |
"n_head": 12,
|
modeling_gpt.py
CHANGED
@@ -1,7 +1,6 @@
|
|
1 |
# Importing libraries
|
2 |
import torch
|
3 |
import torch.nn as nn
|
4 |
-
from typing import Generator
|
5 |
import torch.nn.functional as F
|
6 |
from transformers import PreTrainedModel
|
7 |
from .configuration_gpt import GPTConfig
|
@@ -250,7 +249,7 @@ class GPTModelForTextGeneration(PreTrainedModel):
|
|
250 |
# Autoregressive token generation loop
|
251 |
while tokens.size(1) < max_length:
|
252 |
logits = self.forward(tokens)["logits"][:, -1, :]
|
253 |
-
logits = logits / temperature
|
254 |
|
255 |
if do_sample:
|
256 |
top_k = min(top_k, logits.size(-1)) # Safety check
|
|
|
1 |
# Importing libraries
|
2 |
import torch
|
3 |
import torch.nn as nn
|
|
|
4 |
import torch.nn.functional as F
|
5 |
from transformers import PreTrainedModel
|
6 |
from .configuration_gpt import GPTConfig
|
|
|
249 |
# Autoregressive token generation loop
|
250 |
while tokens.size(1) < max_length:
|
251 |
logits = self.forward(tokens)["logits"][:, -1, :]
|
252 |
+
logits = logits / max(0.01, temperature)
|
253 |
|
254 |
if do_sample:
|
255 |
top_k = min(top_k, logits.size(-1)) # Safety check
|