Hugging Face's logo

zeeshan73
/
mistral_7b_cosine_w_restart_lr

PEFT
TensorBoard
Safetensors
trl
sft
Generated from Trainer
Model card Files Metrics Community
zeeshan73 commited on
Commit
37755ec
·
verified ·
1 Parent(s): 8d4cd0d

Model save

Browse files
Files changed (2) hide show
  1. README.md +96 -0
  2. adapter_model.safetensors +1 -1
README.md ADDED
@@ -0,0 +1,96 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ base_model: mistralai/Mistral-7B-Instruct-v0.3
3
+ datasets:
4
+ - generator
5
+ library_name: peft
6
+ license: apache-2.0
7
+ tags:
8
+ - trl
9
+ - sft
10
+ - generated_from_trainer
11
+ model-index:
12
+ - name: mistral_7b_cosine_lr
13
+ results: []
14
+ ---
15
+
16
+ <!-- This model card has been generated automatically according to the information the Trainer had access to. You
17
+ should probably proofread and complete it, then remove this comment. -->
18
+
19
+ # mistral_7b_cosine_lr
20
+
21
+ This model is a fine-tuned version of [mistralai/Mistral-7B-Instruct-v0.3](https://huggingface.co/mistralai/Mistral-7B-Instruct-v0.3) on the generator dataset.
22
+ It achieves the following results on the evaluation set:
23
+ - Loss: 0.3810
24
+
25
+ ## Model description
26
+
27
+ More information needed
28
+
29
+ ## Intended uses & limitations
30
+
31
+ More information needed
32
+
33
+ ## Training and evaluation data
34
+
35
+ More information needed
36
+
37
+ ## Training procedure
38
+
39
+ ### Training hyperparameters
40
+
41
+ The following hyperparameters were used during training:
42
+ - learning_rate: 0.0002
43
+ - train_batch_size: 2
44
+ - eval_batch_size: 8
45
+ - seed: 42
46
+ - gradient_accumulation_steps: 8
47
+ - total_train_batch_size: 16
48
+ - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
49
+ - lr_scheduler_type: cosine
50
+ - lr_scheduler_warmup_ratio: 0.03
51
+ - lr_scheduler_warmup_steps: 15
52
+ - num_epochs: 3
53
+
54
+ ### Training results
55
+
56
+ | Training Loss | Epoch | Step | Validation Loss |
57
+ |:-------------:|:------:|:----:|:---------------:|
58
+ | 0.8199 | 0.0366 | 10 | 0.6013 |
59
+ | 0.5795 | 0.0732 | 20 | 0.5157 |
60
+ | 0.5089 | 0.1098 | 30 | 0.4825 |
61
+ | 0.4742 | 0.1465 | 40 | 0.4628 |
62
+ | 0.468 | 0.1831 | 50 | 0.4502 |
63
+ | 0.4489 | 0.2197 | 60 | 0.4424 |
64
+ | 0.4471 | 0.2563 | 70 | 0.4378 |
65
+ | 0.4529 | 0.2929 | 80 | 0.4365 |
66
+ | 0.4382 | 0.3295 | 90 | 0.4289 |
67
+ | 0.4352 | 0.3661 | 100 | 0.4267 |
68
+ | 0.4331 | 0.4027 | 110 | 0.4214 |
69
+ | 0.4368 | 0.4394 | 120 | 0.4326 |
70
+ | 0.4244 | 0.4760 | 130 | 0.4188 |
71
+ | 0.4231 | 0.5126 | 140 | 0.4143 |
72
+ | 0.418 | 0.5492 | 150 | 0.4103 |
73
+ | 0.4118 | 0.5858 | 160 | 0.4066 |
74
+ | 0.4151 | 0.6224 | 170 | 0.4047 |
75
+ | 0.4093 | 0.6590 | 180 | 0.4020 |
76
+ | 0.4066 | 0.6957 | 190 | 0.4004 |
77
+ | 0.4072 | 0.7323 | 200 | 0.3970 |
78
+ | 0.4002 | 0.7689 | 210 | 0.3933 |
79
+ | 0.3968 | 0.8055 | 220 | 0.3919 |
80
+ | 0.3967 | 0.8421 | 230 | 0.3893 |
81
+ | 0.3901 | 0.8787 | 240 | 0.3870 |
82
+ | 0.3895 | 0.9153 | 250 | 0.3852 |
83
+ | 0.3929 | 0.9519 | 260 | 0.3823 |
84
+ | 0.3854 | 0.9886 | 270 | 0.3808 |
85
+ | 0.3524 | 1.0252 | 280 | 0.3833 |
86
+ | 0.332 | 1.0618 | 290 | 0.3821 |
87
+ | 0.3283 | 1.0984 | 300 | 0.3810 |
88
+
89
+
90
+ ### Framework versions
91
+
92
+ - PEFT 0.13.2
93
+ - Transformers 4.45.2
94
+ - Pytorch 2.4.1+cu121
95
+ - Datasets 3.0.1
96
+ - Tokenizers 0.20.0
adapter_model.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:17cd69d681315a9be6e7851c9404f23bee063ada55a828690729922eda837543
3
  size 3221320120
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:f0265a07e8f4eaeaf4106ad5c90630e89e545e4b733c243eb437e1f0c94d9e10
3
  size 3221320120