Update README.md
Browse files
README.md
CHANGED
@@ -5,12 +5,62 @@ tags:
|
|
5 |
- axolotl
|
6 |
- generated_from_trainer
|
7 |
model-index:
|
8 |
-
- name: Llama-3-8B-
|
9 |
results: []
|
10 |
---
|
11 |
|
12 |
-
|
13 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
14 |
|
15 |
[<img src="https://raw.githubusercontent.com/OpenAccess-AI-Collective/axolotl/main/image/axolotl-badge-web.png" alt="Built with Axolotl" width="200" height="32"/>](https://github.com/OpenAccess-AI-Collective/axolotl)
|
16 |
<details><summary>See axolotl config</summary>
|
@@ -26,25 +76,18 @@ load_in_4bit: false
|
|
26 |
strict: false
|
27 |
|
28 |
datasets:
|
29 |
-
- path:
|
30 |
type: sharegpt
|
31 |
conversation: llama3
|
32 |
dataset_prepared_path: last_run_prepared
|
33 |
val_set_size: 0.001
|
34 |
-
output_dir: ./out_Llama-3-
|
35 |
|
36 |
sequence_len: 8192
|
37 |
sample_packing: true
|
38 |
eval_sample_packing: false
|
39 |
pad_to_sequence_len: true
|
40 |
|
41 |
-
wandb_project: SynDa
|
42 |
-
wandb_entity:
|
43 |
-
wandb_watch:
|
44 |
-
wandb_name: Llama-3-70B-SynDa-100K-FilteredL-2EP-FFT
|
45 |
-
wandb_log_model:
|
46 |
-
hub_model_id: SynDa/Llama-3-8B-SynDa-70BQA-100K-Filtered-L
|
47 |
-
|
48 |
gradient_accumulation_steps: 8
|
49 |
micro_batch_size: 1
|
50 |
num_epochs: 2
|
@@ -82,59 +125,3 @@ special_tokens:
|
|
82 |
```
|
83 |
|
84 |
</details><br>
|
85 |
-
|
86 |
-
# Llama-3-8B-SynDa-70BQA-100K-Filtered-L
|
87 |
-
|
88 |
-
This model is a fine-tuned version of [meta-llama/Meta-Llama-3-8B](https://huggingface.co/meta-llama/Meta-Llama-3-8B) on the None dataset.
|
89 |
-
It achieves the following results on the evaluation set:
|
90 |
-
- Loss: 0.5056
|
91 |
-
|
92 |
-
## Model description
|
93 |
-
|
94 |
-
More information needed
|
95 |
-
|
96 |
-
## Intended uses & limitations
|
97 |
-
|
98 |
-
More information needed
|
99 |
-
|
100 |
-
## Training and evaluation data
|
101 |
-
|
102 |
-
More information needed
|
103 |
-
|
104 |
-
## Training procedure
|
105 |
-
|
106 |
-
### Training hyperparameters
|
107 |
-
|
108 |
-
The following hyperparameters were used during training:
|
109 |
-
- learning_rate: 2e-05
|
110 |
-
- train_batch_size: 1
|
111 |
-
- eval_batch_size: 1
|
112 |
-
- seed: 42
|
113 |
-
- distributed_type: multi-GPU
|
114 |
-
- num_devices: 4
|
115 |
-
- gradient_accumulation_steps: 8
|
116 |
-
- total_train_batch_size: 32
|
117 |
-
- total_eval_batch_size: 4
|
118 |
-
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
|
119 |
-
- lr_scheduler_type: cosine
|
120 |
-
- lr_scheduler_warmup_steps: 100
|
121 |
-
- num_epochs: 2
|
122 |
-
|
123 |
-
### Training results
|
124 |
-
|
125 |
-
| Training Loss | Epoch | Step | Validation Loss |
|
126 |
-
|:-------------:|:------:|:----:|:---------------:|
|
127 |
-
| 0.8869 | 0.0036 | 1 | 0.9139 |
|
128 |
-
| 0.5854 | 0.3344 | 92 | 0.6158 |
|
129 |
-
| 0.5218 | 0.6688 | 184 | 0.5455 |
|
130 |
-
| 0.4878 | 1.0032 | 276 | 0.5125 |
|
131 |
-
| 0.3734 | 1.3226 | 368 | 0.5091 |
|
132 |
-
| 0.3647 | 1.6570 | 460 | 0.5056 |
|
133 |
-
|
134 |
-
|
135 |
-
### Framework versions
|
136 |
-
|
137 |
-
- Transformers 4.40.2
|
138 |
-
- Pytorch 2.3.0+cu121
|
139 |
-
- Datasets 2.19.1
|
140 |
-
- Tokenizers 0.19.1
|
|
|
5 |
- axolotl
|
6 |
- generated_from_trainer
|
7 |
model-index:
|
8 |
+
- name: Llama-3-8B-Magpie-Pro-SFT-100K-v0.1
|
9 |
results: []
|
10 |
---
|
11 |
|
12 |
+
|
13 |
+
# Magpie-Align/Llama-3-8B-Magpie-Pro-SFT-100K-v0.1
|
14 |
+
|
15 |
+
Project Web: [https://magpie-align.github.io/](https://magpie-align.github.io/)
|
16 |
+
|
17 |
+
Arxiv Technical Report: [https://arxiv.org/abs/2406.08464](https://arxiv.org/abs/2406.08464)
|
18 |
+
|
19 |
+
Codes: [https://github.com/magpie-align/magpie](https://github.com/magpie-align/magpie)
|
20 |
+
|
21 |
+
## About This Model
|
22 |
+
|
23 |
+
This model is a fine-tuned version of [meta-llama/Meta-Llama-3-8B](https://huggingface.co/meta-llama/Meta-Llama-3-8B) on **First 100K data** of [Magpie-Align/Magpie-Pro-300K-Filtered](https://huggingface.co/datasets/Magpie-Align/Magpie-Pro-300K-Filtered) dataset.
|
24 |
+
|
25 |
+
Please use [Magpie-Align/Llama-3-8B-Magpie-Pro-SFT-v0.1](https://huggingface.co/Magpie-Align/Llama-3-8B-Magpie-Pro-SFT-v0.1) with better performance.
|
26 |
+
|
27 |
+
## Training procedure
|
28 |
+
|
29 |
+
### Training hyperparameters
|
30 |
+
|
31 |
+
The following hyperparameters were used during training:
|
32 |
+
- learning_rate: 2e-05
|
33 |
+
- train_batch_size: 1
|
34 |
+
- eval_batch_size: 1
|
35 |
+
- seed: 42
|
36 |
+
- distributed_type: multi-GPU
|
37 |
+
- num_devices: 4
|
38 |
+
- gradient_accumulation_steps: 8
|
39 |
+
- total_train_batch_size: 32
|
40 |
+
- total_eval_batch_size: 4
|
41 |
+
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
|
42 |
+
- lr_scheduler_type: cosine
|
43 |
+
- lr_scheduler_warmup_steps: 100
|
44 |
+
- num_epochs: 2
|
45 |
+
|
46 |
+
### Training results
|
47 |
+
|
48 |
+
| Training Loss | Epoch | Step | Validation Loss |
|
49 |
+
|:-------------:|:------:|:----:|:---------------:|
|
50 |
+
| 0.8869 | 0.0036 | 1 | 0.9139 |
|
51 |
+
| 0.5854 | 0.3344 | 92 | 0.6158 |
|
52 |
+
| 0.5218 | 0.6688 | 184 | 0.5455 |
|
53 |
+
| 0.4878 | 1.0032 | 276 | 0.5125 |
|
54 |
+
| 0.3734 | 1.3226 | 368 | 0.5091 |
|
55 |
+
| 0.3647 | 1.6570 | 460 | 0.5056 |
|
56 |
+
|
57 |
+
|
58 |
+
### Framework versions
|
59 |
+
|
60 |
+
- Transformers 4.40.2
|
61 |
+
- Pytorch 2.3.0+cu121
|
62 |
+
- Datasets 2.19.1
|
63 |
+
- Tokenizers 0.19.1
|
64 |
|
65 |
[<img src="https://raw.githubusercontent.com/OpenAccess-AI-Collective/axolotl/main/image/axolotl-badge-web.png" alt="Built with Axolotl" width="200" height="32"/>](https://github.com/OpenAccess-AI-Collective/axolotl)
|
66 |
<details><summary>See axolotl config</summary>
|
|
|
76 |
strict: false
|
77 |
|
78 |
datasets:
|
79 |
+
- path: Magpie-Align/Magpie-Pro-300K-Filtered-First100K
|
80 |
type: sharegpt
|
81 |
conversation: llama3
|
82 |
dataset_prepared_path: last_run_prepared
|
83 |
val_set_size: 0.001
|
84 |
+
output_dir: ./out_Llama-3-8B-Magpie-Pro-100K-FilteredL
|
85 |
|
86 |
sequence_len: 8192
|
87 |
sample_packing: true
|
88 |
eval_sample_packing: false
|
89 |
pad_to_sequence_len: true
|
90 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
91 |
gradient_accumulation_steps: 8
|
92 |
micro_batch_size: 1
|
93 |
num_epochs: 2
|
|
|
125 |
```
|
126 |
|
127 |
</details><br>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|