Hugging Face's logo

Grigorij
/
mistral_instruct_generation

PEFT
Safetensors
trl
sft
Generated from Trainer
Model card Files
xet
Community
Grigorij commited on
Commit
e0bbcdb
·
1 Parent(s): 5bb1d8a

Model save

Browse files
Files changed (1) hide show
  1. README.md +17 -19
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 [mistralai/Mistral-7B-Instruct-v0.2](https://huggingface.co/mistralai/Mistral-7B-Instruct-v0.2) on an unknown dataset.
20
  It achieves the following results on the evaluation set:
21
- - Loss: 0.7197
22
 
23
  ## Model description
24
 
@@ -44,29 +44,27 @@ The following hyperparameters were used during training:
44
  - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
45
  - lr_scheduler_type: constant
46
  - lr_scheduler_warmup_steps: 0.03
47
- - training_steps: 170
48
 
49
  ### Training results
50
 
51
  | Training Loss | Epoch | Step | Validation Loss |
52
  |:-------------:|:-----:|:----:|:---------------:|
53
- | 2.784 | 0.4 | 10 | 2.0951 |
54
- | 1.6947 | 0.8 | 20 | 1.2317 |
55
- | 1.0477 | 1.2 | 30 | 0.9772 |
56
- | 0.9452 | 1.6 | 40 | 0.9071 |
57
- | 0.9092 | 2.0 | 50 | 0.8293 |
58
- | 0.7453 | 2.4 | 60 | 0.7538 |
59
- | 0.7201 | 2.8 | 70 | 0.6689 |
60
- | 0.5659 | 3.2 | 80 | 0.6524 |
61
- | 0.5609 | 3.6 | 90 | 0.6494 |
62
- | 0.6039 | 4.0 | 100 | 0.6349 |
63
- | 0.5073 | 4.4 | 110 | 0.6542 |
64
- | 0.4997 | 4.8 | 120 | 0.6404 |
65
- | 0.495 | 5.2 | 130 | 0.6557 |
66
- | 0.453 | 5.6 | 140 | 0.6815 |
67
- | 0.4803 | 6.0 | 150 | 0.6361 |
68
- | 0.4233 | 6.4 | 160 | 0.7303 |
69
- | 0.4209 | 6.8 | 170 | 0.7197 |
70
 
71
 
72
  ### Framework versions
 
18
 
19
  This model is a fine-tuned version of [mistralai/Mistral-7B-Instruct-v0.2](https://huggingface.co/mistralai/Mistral-7B-Instruct-v0.2) on an unknown dataset.
20
  It achieves the following results on the evaluation set:
21
+ - Loss: 0.5788
22
 
23
  ## Model description
24
 
 
44
  - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
45
  - lr_scheduler_type: constant
46
  - lr_scheduler_warmup_steps: 0.03
47
+ - training_steps: 150
48
 
49
  ### Training results
50
 
51
  | Training Loss | Epoch | Step | Validation Loss |
52
  |:-------------:|:-----:|:----:|:---------------:|
53
+ | 4.0261 | 0.4 | 10 | 2.8697 |
54
+ | 2.2616 | 0.8 | 20 | 1.6009 |
55
+ | 1.1871 | 1.2 | 30 | 0.9599 |
56
+ | 0.8522 | 1.6 | 40 | 0.7228 |
57
+ | 0.7375 | 2.0 | 50 | 0.6601 |
58
+ | 0.5916 | 2.4 | 60 | 0.6184 |
59
+ | 0.6219 | 2.8 | 70 | 0.5957 |
60
+ | 0.5025 | 3.2 | 80 | 0.5980 |
61
+ | 0.5148 | 3.6 | 90 | 0.5849 |
62
+ | 0.5502 | 4.0 | 100 | 0.5639 |
63
+ | 0.4414 | 4.4 | 110 | 0.5875 |
64
+ | 0.4423 | 4.8 | 120 | 0.5847 |
65
+ | 0.43 | 5.2 | 130 | 0.5902 |
66
+ | 0.3843 | 5.6 | 140 | 0.6223 |
67
+ | 0.4173 | 6.0 | 150 | 0.5788 |
 
 
68
 
69
 
70
  ### Framework versions