Hugging Face's logo

DLingo
/
qwen2-2b-instruct-trl-sft-mrg

PEFT
Safetensors
qwen2_vl
trl
sft
Generated from Trainer
Model card Files Community
DLingo commited on
Commit
4037c2b
·
verified ·
1 Parent(s): 5ccb2ab

Model save

Browse files
Files changed (1) hide show
  1. README.md +31 -31
README.md CHANGED
@@ -18,7 +18,7 @@ should probably proofread and complete it, then remove this comment. -->
18
 
19
  This model is a fine-tuned version of [Qwen/Qwen2-VL-2B-Instruct](https://huggingface.co/Qwen/Qwen2-VL-2B-Instruct) on an unknown dataset.
20
  It achieves the following results on the evaluation set:
21
- - Loss: 0.6486
22
 
23
  ## Model description
24
 
@@ -52,36 +52,36 @@ The following hyperparameters were used during training:
52
 
53
  | Training Loss | Epoch | Step | Validation Loss |
54
  |:-------------:|:-------:|:----:|:---------------:|
55
- | 3.1914 | 0.4942 | 32 | 3.1004 |
56
- | 2.6694 | 0.9884 | 64 | 2.3834 |
57
- | 1.6251 | 1.4846 | 96 | 1.3020 |
58
- | 1.2153 | 1.9788 | 128 | 1.0890 |
59
- | 1.02 | 2.4749 | 160 | 0.9552 |
60
- | 0.8604 | 2.9691 | 192 | 0.8315 |
61
- | 0.8306 | 3.4653 | 224 | 0.7960 |
62
- | 0.7928 | 3.9595 | 256 | 0.7662 |
63
- | 0.8038 | 4.4556 | 288 | 0.7454 |
64
- | 0.7275 | 4.9498 | 320 | 0.7290 |
65
- | 0.7229 | 5.4459 | 352 | 0.7150 |
66
- | 0.707 | 5.9402 | 384 | 0.7046 |
67
- | 0.6546 | 6.4363 | 416 | 0.6962 |
68
- | 0.6852 | 6.9305 | 448 | 0.6866 |
69
- | 0.6646 | 7.4266 | 480 | 0.6793 |
70
- | 0.6336 | 7.9208 | 512 | 0.6724 |
71
- | 0.674 | 8.4170 | 544 | 0.6680 |
72
- | 0.6936 | 8.9112 | 576 | 0.6646 |
73
- | 0.6468 | 9.4073 | 608 | 0.6627 |
74
- | 0.6208 | 9.9015 | 640 | 0.6574 |
75
- | 0.6236 | 10.3977 | 672 | 0.6555 |
76
- | 0.6962 | 10.8919 | 704 | 0.6533 |
77
- | 0.6333 | 11.3880 | 736 | 0.6521 |
78
- | 0.625 | 11.8822 | 768 | 0.6501 |
79
- | 0.6226 | 12.3784 | 800 | 0.6511 |
80
- | 0.6543 | 12.8726 | 832 | 0.6496 |
81
- | 0.6582 | 13.3687 | 864 | 0.6492 |
82
- | 0.6528 | 13.8629 | 896 | 0.6486 |
83
- | 0.6209 | 14.3591 | 928 | 0.6485 |
84
- | 0.6067 | 14.8533 | 960 | 0.6486 |
85
 
86
 
87
  ### Framework versions
 
18
 
19
  This model is a fine-tuned version of [Qwen/Qwen2-VL-2B-Instruct](https://huggingface.co/Qwen/Qwen2-VL-2B-Instruct) on an unknown dataset.
20
  It achieves the following results on the evaluation set:
21
+ - Loss: 1.2568
22
 
23
  ## Model description
24
 
 
52
 
53
  | Training Loss | Epoch | Step | Validation Loss |
54
  |:-------------:|:-------:|:----:|:---------------:|
55
+ | 3.0236 | 0.4942 | 32 | 2.9521 |
56
+ | 2.6642 | 0.9884 | 64 | 2.4341 |
57
+ | 2.156 | 1.4846 | 96 | 1.9677 |
58
+ | 1.9011 | 1.9788 | 128 | 1.7312 |
59
+ | 1.6955 | 2.4749 | 160 | 1.6093 |
60
+ | 1.5552 | 2.9691 | 192 | 1.5437 |
61
+ | 1.5361 | 3.4653 | 224 | 1.4991 |
62
+ | 1.4831 | 3.9595 | 256 | 1.4554 |
63
+ | 1.5036 | 4.4556 | 288 | 1.4261 |
64
+ | 1.3815 | 4.9498 | 320 | 1.3991 |
65
+ | 1.3762 | 5.4459 | 352 | 1.3760 |
66
+ | 1.3636 | 5.9402 | 384 | 1.3562 |
67
+ | 1.2826 | 6.4363 | 416 | 1.3424 |
68
+ | 1.3178 | 6.9305 | 448 | 1.3256 |
69
+ | 1.2689 | 7.4266 | 480 | 1.3123 |
70
+ | 1.2163 | 7.9208 | 512 | 1.3019 |
71
+ | 1.284 | 8.4170 | 544 | 1.2920 |
72
+ | 1.3356 | 8.9112 | 576 | 1.2862 |
73
+ | 1.2359 | 9.4073 | 608 | 1.2820 |
74
+ | 1.2157 | 9.9015 | 640 | 1.2746 |
75
+ | 1.1936 | 10.3977 | 672 | 1.2709 |
76
+ | 1.3181 | 10.8919 | 704 | 1.2659 |
77
+ | 1.2266 | 11.3880 | 736 | 1.2641 |
78
+ | 1.213 | 11.8822 | 768 | 1.2605 |
79
+ | 1.1997 | 12.3784 | 800 | 1.2603 |
80
+ | 1.2584 | 12.8726 | 832 | 1.2577 |
81
+ | 1.2547 | 13.3687 | 864 | 1.2576 |
82
+ | 1.2544 | 13.8629 | 896 | 1.2574 |
83
+ | 1.203 | 14.3591 | 928 | 1.2569 |
84
+ | 1.1467 | 14.8533 | 960 | 1.2568 |
85
 
86
 
87
  ### Framework versions