Model save
Browse files
README.md
ADDED
@@ -0,0 +1,100 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
---
|
2 |
+
license: llama3
|
3 |
+
library_name: peft
|
4 |
+
tags:
|
5 |
+
- trl
|
6 |
+
- sft
|
7 |
+
- generated_from_trainer
|
8 |
+
base_model: meta-llama/Meta-Llama-3-8B-Instruct
|
9 |
+
datasets:
|
10 |
+
- generator
|
11 |
+
model-index:
|
12 |
+
- name: cls_alldata_llama3_v1
|
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 |
+
# cls_alldata_llama3_v1
|
20 |
+
|
21 |
+
This model is a fine-tuned version of [meta-llama/Meta-Llama-3-8B-Instruct](https://huggingface.co/meta-llama/Meta-Llama-3-8B-Instruct) on the generator dataset.
|
22 |
+
It achieves the following results on the evaluation set:
|
23 |
+
- Loss: 0.4523
|
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: 4
|
47 |
+
- total_train_batch_size: 8
|
48 |
+
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
|
49 |
+
- lr_scheduler_type: constant
|
50 |
+
- lr_scheduler_warmup_ratio: 0.03
|
51 |
+
- num_epochs: 2
|
52 |
+
- mixed_precision_training: Native AMP
|
53 |
+
|
54 |
+
### Training results
|
55 |
+
|
56 |
+
| Training Loss | Epoch | Step | Validation Loss |
|
57 |
+
|:-------------:|:------:|:----:|:---------------:|
|
58 |
+
| 0.6921 | 0.0582 | 20 | 0.6831 |
|
59 |
+
| 0.5975 | 0.1164 | 40 | 0.6416 |
|
60 |
+
| 0.6107 | 0.1747 | 60 | 0.6082 |
|
61 |
+
| 0.5609 | 0.2329 | 80 | 0.5883 |
|
62 |
+
| 0.5857 | 0.2911 | 100 | 0.5761 |
|
63 |
+
| 0.5386 | 0.3493 | 120 | 0.5660 |
|
64 |
+
| 0.5176 | 0.4076 | 140 | 0.5529 |
|
65 |
+
| 0.5317 | 0.4658 | 160 | 0.5379 |
|
66 |
+
| 0.5244 | 0.5240 | 180 | 0.5292 |
|
67 |
+
| 0.5218 | 0.5822 | 200 | 0.5234 |
|
68 |
+
| 0.5003 | 0.6405 | 220 | 0.5207 |
|
69 |
+
| 0.5024 | 0.6987 | 240 | 0.5096 |
|
70 |
+
| 0.4913 | 0.7569 | 260 | 0.5062 |
|
71 |
+
| 0.5174 | 0.8151 | 280 | 0.5003 |
|
72 |
+
| 0.4675 | 0.8734 | 300 | 0.4968 |
|
73 |
+
| 0.5137 | 0.9316 | 320 | 0.4903 |
|
74 |
+
| 0.4883 | 0.9898 | 340 | 0.4869 |
|
75 |
+
| 0.3616 | 1.0480 | 360 | 0.4935 |
|
76 |
+
| 0.3713 | 1.1063 | 380 | 0.4890 |
|
77 |
+
| 0.365 | 1.1645 | 400 | 0.4856 |
|
78 |
+
| 0.3732 | 1.2227 | 420 | 0.4838 |
|
79 |
+
| 0.3717 | 1.2809 | 440 | 0.4842 |
|
80 |
+
| 0.3657 | 1.3392 | 460 | 0.4811 |
|
81 |
+
| 0.3767 | 1.3974 | 480 | 0.4762 |
|
82 |
+
| 0.3859 | 1.4556 | 500 | 0.4763 |
|
83 |
+
| 0.3773 | 1.5138 | 520 | 0.4712 |
|
84 |
+
| 0.3615 | 1.5721 | 540 | 0.4671 |
|
85 |
+
| 0.3656 | 1.6303 | 560 | 0.4666 |
|
86 |
+
| 0.3497 | 1.6885 | 580 | 0.4658 |
|
87 |
+
| 0.3818 | 1.7467 | 600 | 0.4621 |
|
88 |
+
| 0.3759 | 1.8049 | 620 | 0.4626 |
|
89 |
+
| 0.3539 | 1.8632 | 640 | 0.4551 |
|
90 |
+
| 0.3985 | 1.9214 | 660 | 0.4525 |
|
91 |
+
| 0.3668 | 1.9796 | 680 | 0.4523 |
|
92 |
+
|
93 |
+
|
94 |
+
### Framework versions
|
95 |
+
|
96 |
+
- PEFT 0.11.1
|
97 |
+
- Transformers 4.41.1
|
98 |
+
- Pytorch 2.3.0+cu121
|
99 |
+
- Datasets 2.19.1
|
100 |
+
- Tokenizers 0.19.1
|