GradientGuru
commited on
Commit
•
2026f61
1
Parent(s):
fc64079
Update modeling_baichuan.py
Browse files- modeling_baichuan.py +8 -1
modeling_baichuan.py
CHANGED
@@ -4,6 +4,7 @@ import math
|
|
4 |
from typing import List, Optional, Tuple, Union
|
5 |
|
6 |
import torch
|
|
|
7 |
from torch.nn import CrossEntropyLoss
|
8 |
from transformers import PreTrainedModel
|
9 |
from transformers.activations import ACT2FN
|
@@ -235,7 +236,13 @@ class BaichuanModel(BaichuanPreTrainedModel):
|
|
235 |
self.post_init()
|
236 |
self.max_cache_pos = config.model_max_length
|
237 |
self.first_run = True
|
238 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
239 |
def get_alibi_mask(self, tensor, seq_length_with_past):
|
240 |
if self.first_run:
|
241 |
self.first_run = False
|
|
|
4 |
from typing import List, Optional, Tuple, Union
|
5 |
|
6 |
import torch
|
7 |
+
import torch.utils.checkpoint
|
8 |
from torch.nn import CrossEntropyLoss
|
9 |
from transformers import PreTrainedModel
|
10 |
from transformers.activations import ACT2FN
|
|
|
236 |
self.post_init()
|
237 |
self.max_cache_pos = config.model_max_length
|
238 |
self.first_run = True
|
239 |
+
|
240 |
+
def get_input_embeddings(self):
|
241 |
+
return self.embed_tokens
|
242 |
+
|
243 |
+
def set_input_embeddings(self, value):
|
244 |
+
self.embed_tokens = value
|
245 |
+
|
246 |
def get_alibi_mask(self, tensor, seq_length_with_past):
|
247 |
if self.first_run:
|
248 |
self.first_run = False
|