yunaseo/google_gemma_lora_emotion_detection
Browse files- README.md +35 -57
- adapter_config.json +35 -0
- adapter_model.safetensors +3 -0
- training_args.bin +1 -1
README.md
CHANGED
@@ -1,8 +1,9 @@
|
|
1 |
---
|
2 |
license: gemma
|
3 |
-
|
4 |
tags:
|
5 |
- generated_from_trainer
|
|
|
6 |
metrics:
|
7 |
- accuracy
|
8 |
model-index:
|
@@ -17,10 +18,10 @@ should probably proofread and complete it, then remove this comment. -->
|
|
17 |
|
18 |
This model is a fine-tuned version of [google/gemma-1.1-2b-it](https://huggingface.co/google/gemma-1.1-2b-it) on an unknown dataset.
|
19 |
It achieves the following results on the evaluation set:
|
20 |
-
- Loss: 0.
|
21 |
-
- F1 Micro: 0.
|
22 |
-
- F1 Macro: 0.
|
23 |
-
- Accuracy: 0.
|
24 |
|
25 |
## Model description
|
26 |
|
@@ -40,11 +41,11 @@ More information needed
|
|
40 |
|
41 |
The following hyperparameters were used during training:
|
42 |
- learning_rate: 0.0001
|
43 |
-
- train_batch_size:
|
44 |
-
- eval_batch_size:
|
45 |
- seed: 42
|
46 |
- gradient_accumulation_steps: 4
|
47 |
-
- total_train_batch_size:
|
48 |
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
|
49 |
- lr_scheduler_type: linear
|
50 |
- num_epochs: 5
|
@@ -53,59 +54,36 @@ The following hyperparameters were used during training:
|
|
53 |
|
54 |
| Training Loss | Epoch | Step | Validation Loss | F1 Micro | F1 Macro | Accuracy |
|
55 |
|:-------------:|:------:|:----:|:---------------:|:--------:|:--------:|:--------:|
|
56 |
-
| 0.
|
57 |
-
| 0.
|
58 |
-
| 0.
|
59 |
-
| 0.
|
60 |
-
| 0.
|
61 |
-
| 0.
|
62 |
-
| 0.
|
63 |
-
| 0.
|
64 |
-
| 0.
|
65 |
-
| 0.
|
66 |
-
| 0.
|
67 |
-
| 0.
|
68 |
-
| 0.
|
69 |
-
| 0.
|
70 |
-
| 0.
|
71 |
-
| 0.
|
72 |
-
| 0.
|
73 |
-
| 0.
|
74 |
-
| 0.
|
75 |
-
| 0.
|
76 |
-
| 0.
|
77 |
-
| 0.
|
78 |
-
| 0.
|
79 |
-
| 0.
|
80 |
-
| 0.14 | 2.5873 | 500 | 0.6777 | 0.6855 | 0.5826 | 0.2104 |
|
81 |
-
| 0.146 | 2.6908 | 520 | 0.6699 | 0.6837 | 0.5840 | 0.2129 |
|
82 |
-
| 0.138 | 2.7943 | 540 | 0.6954 | 0.6884 | 0.5820 | 0.2369 |
|
83 |
-
| 0.1302 | 2.8978 | 560 | 0.7090 | 0.6828 | 0.5777 | 0.2220 |
|
84 |
-
| 0.1324 | 3.0013 | 580 | 0.7075 | 0.6845 | 0.5818 | 0.2259 |
|
85 |
-
| 0.0472 | 3.1048 | 600 | 0.8346 | 0.6867 | 0.5575 | 0.2414 |
|
86 |
-
| 0.0544 | 3.2083 | 620 | 0.7725 | 0.6785 | 0.5706 | 0.2207 |
|
87 |
-
| 0.0483 | 3.3118 | 640 | 0.8136 | 0.6865 | 0.5659 | 0.2291 |
|
88 |
-
| 0.0465 | 3.4153 | 660 | 0.8333 | 0.6797 | 0.5613 | 0.2278 |
|
89 |
-
| 0.0511 | 3.5188 | 680 | 0.8234 | 0.6852 | 0.5641 | 0.2265 |
|
90 |
-
| 0.0511 | 3.6223 | 700 | 0.8298 | 0.6905 | 0.5712 | 0.2401 |
|
91 |
-
| 0.0406 | 3.7257 | 720 | 0.8292 | 0.6886 | 0.5721 | 0.2421 |
|
92 |
-
| 0.0565 | 3.8292 | 740 | 0.8266 | 0.6927 | 0.5721 | 0.2408 |
|
93 |
-
| 0.0554 | 3.9327 | 760 | 0.7764 | 0.6887 | 0.5765 | 0.2350 |
|
94 |
-
| 0.0319 | 4.0362 | 780 | 0.8450 | 0.6825 | 0.5650 | 0.2388 |
|
95 |
-
| 0.0161 | 4.1397 | 800 | 0.8948 | 0.6892 | 0.5648 | 0.2524 |
|
96 |
-
| 0.0174 | 4.2432 | 820 | 0.9146 | 0.6910 | 0.5659 | 0.2570 |
|
97 |
-
| 0.0168 | 4.3467 | 840 | 0.9068 | 0.6874 | 0.5657 | 0.2414 |
|
98 |
-
| 0.0184 | 4.4502 | 860 | 0.9225 | 0.6872 | 0.5615 | 0.2531 |
|
99 |
-
| 0.0123 | 4.5537 | 880 | 0.9062 | 0.6882 | 0.5639 | 0.2511 |
|
100 |
-
| 0.0149 | 4.6572 | 900 | 0.9087 | 0.6889 | 0.5660 | 0.2492 |
|
101 |
-
| 0.0199 | 4.7607 | 920 | 0.8948 | 0.6917 | 0.5722 | 0.2472 |
|
102 |
-
| 0.0144 | 4.8642 | 940 | 0.8944 | 0.6929 | 0.5724 | 0.2518 |
|
103 |
-
| 0.015 | 4.9677 | 960 | 0.8963 | 0.6925 | 0.5709 | 0.2531 |
|
104 |
|
105 |
|
106 |
### Framework versions
|
107 |
|
|
|
108 |
- Transformers 4.40.2
|
109 |
- Pytorch 2.2.1+cu121
|
110 |
- Datasets 2.19.1
|
111 |
-
- Tokenizers 0.19.1
|
|
|
1 |
---
|
2 |
license: gemma
|
3 |
+
library_name: peft
|
4 |
tags:
|
5 |
- generated_from_trainer
|
6 |
+
base_model: google/gemma-1.1-2b-it
|
7 |
metrics:
|
8 |
- accuracy
|
9 |
model-index:
|
|
|
18 |
|
19 |
This model is a fine-tuned version of [google/gemma-1.1-2b-it](https://huggingface.co/google/gemma-1.1-2b-it) on an unknown dataset.
|
20 |
It achieves the following results on the evaluation set:
|
21 |
+
- Loss: 0.4792
|
22 |
+
- F1 Micro: 0.6970
|
23 |
+
- F1 Macro: 0.6089
|
24 |
+
- Accuracy: 0.2104
|
25 |
|
26 |
## Model description
|
27 |
|
|
|
41 |
|
42 |
The following hyperparameters were used during training:
|
43 |
- learning_rate: 0.0001
|
44 |
+
- train_batch_size: 16
|
45 |
+
- eval_batch_size: 16
|
46 |
- seed: 42
|
47 |
- gradient_accumulation_steps: 4
|
48 |
+
- total_train_batch_size: 64
|
49 |
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
|
50 |
- lr_scheduler_type: linear
|
51 |
- num_epochs: 5
|
|
|
54 |
|
55 |
| Training Loss | Epoch | Step | Validation Loss | F1 Micro | F1 Macro | Accuracy |
|
56 |
|:-------------:|:------:|:----:|:---------------:|:--------:|:--------:|:--------:|
|
57 |
+
| 0.7081 | 0.2067 | 20 | 0.6048 | 0.6244 | 0.5113 | 0.1528 |
|
58 |
+
| 0.5228 | 0.4134 | 40 | 0.5096 | 0.6713 | 0.5815 | 0.1883 |
|
59 |
+
| 0.5048 | 0.6202 | 60 | 0.4928 | 0.7002 | 0.5865 | 0.2155 |
|
60 |
+
| 0.5129 | 0.8269 | 80 | 0.4792 | 0.6970 | 0.6089 | 0.2104 |
|
61 |
+
| 0.4842 | 1.0336 | 100 | 0.4801 | 0.6972 | 0.6023 | 0.2369 |
|
62 |
+
| 0.3372 | 1.2403 | 120 | 0.5545 | 0.6687 | 0.5877 | 0.1761 |
|
63 |
+
| 0.3302 | 1.4470 | 140 | 0.5374 | 0.6895 | 0.6020 | 0.2019 |
|
64 |
+
| 0.3342 | 1.6537 | 160 | 0.5330 | 0.6860 | 0.5993 | 0.2117 |
|
65 |
+
| 0.3392 | 1.8605 | 180 | 0.5190 | 0.6894 | 0.5913 | 0.2006 |
|
66 |
+
| 0.2844 | 2.0672 | 200 | 0.5853 | 0.6891 | 0.5819 | 0.2369 |
|
67 |
+
| 0.1458 | 2.2739 | 220 | 0.7038 | 0.6743 | 0.5749 | 0.2097 |
|
68 |
+
| 0.1508 | 2.4806 | 240 | 0.6808 | 0.6802 | 0.5834 | 0.1994 |
|
69 |
+
| 0.1481 | 2.6873 | 260 | 0.7026 | 0.6773 | 0.5721 | 0.2 |
|
70 |
+
| 0.1378 | 2.8941 | 280 | 0.7336 | 0.6790 | 0.5768 | 0.2162 |
|
71 |
+
| 0.0961 | 3.1008 | 300 | 0.8397 | 0.6709 | 0.5465 | 0.2272 |
|
72 |
+
| 0.0552 | 3.3075 | 320 | 0.8260 | 0.6743 | 0.5654 | 0.2168 |
|
73 |
+
| 0.0509 | 3.5142 | 340 | 0.8692 | 0.6777 | 0.5666 | 0.2233 |
|
74 |
+
| 0.0489 | 3.7209 | 360 | 0.8505 | 0.6874 | 0.5722 | 0.2388 |
|
75 |
+
| 0.0526 | 3.9276 | 380 | 0.8269 | 0.6842 | 0.5778 | 0.2233 |
|
76 |
+
| 0.0278 | 4.1344 | 400 | 0.9280 | 0.6813 | 0.5557 | 0.2414 |
|
77 |
+
| 0.0187 | 4.3411 | 420 | 0.9390 | 0.6829 | 0.5588 | 0.2382 |
|
78 |
+
| 0.0169 | 4.5478 | 440 | 0.9510 | 0.6834 | 0.5612 | 0.2485 |
|
79 |
+
| 0.0158 | 4.7545 | 460 | 0.9325 | 0.6819 | 0.5612 | 0.2427 |
|
80 |
+
| 0.0161 | 4.9612 | 480 | 0.9311 | 0.6822 | 0.5634 | 0.2440 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
81 |
|
82 |
|
83 |
### Framework versions
|
84 |
|
85 |
+
- PEFT 0.10.0
|
86 |
- Transformers 4.40.2
|
87 |
- Pytorch 2.2.1+cu121
|
88 |
- Datasets 2.19.1
|
89 |
+
- Tokenizers 0.19.1
|
adapter_config.json
ADDED
@@ -0,0 +1,35 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"alpha_pattern": {},
|
3 |
+
"auto_mapping": null,
|
4 |
+
"base_model_name_or_path": "google/gemma-1.1-2b-it",
|
5 |
+
"bias": "lora_only",
|
6 |
+
"fan_in_fan_out": false,
|
7 |
+
"inference_mode": true,
|
8 |
+
"init_lora_weights": true,
|
9 |
+
"layer_replication": null,
|
10 |
+
"layers_pattern": null,
|
11 |
+
"layers_to_transform": null,
|
12 |
+
"loftq_config": {},
|
13 |
+
"lora_alpha": 256,
|
14 |
+
"lora_dropout": 0.01,
|
15 |
+
"megatron_config": null,
|
16 |
+
"megatron_core": "megatron.core",
|
17 |
+
"modules_to_save": null,
|
18 |
+
"peft_type": "LORA",
|
19 |
+
"r": 128,
|
20 |
+
"rank_pattern": {},
|
21 |
+
"revision": null,
|
22 |
+
"target_modules": [
|
23 |
+
"gate_proj",
|
24 |
+
"q_proj",
|
25 |
+
"v_proj",
|
26 |
+
"down_proj",
|
27 |
+
"up_proj",
|
28 |
+
"score",
|
29 |
+
"o_proj",
|
30 |
+
"k_proj"
|
31 |
+
],
|
32 |
+
"task_type": "SEQ_CLS",
|
33 |
+
"use_dora": false,
|
34 |
+
"use_rslora": false
|
35 |
+
}
|
adapter_model.safetensors
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:1cdd6c5dd3983ce0e322336bfe21c89e94579dfb5ac094a05a5c4ef0a43d33d9
|
3 |
+
size 630860528
|
training_args.bin
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 5112
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:ea96437ebc9da0288a42bc399bddcb8a110fe9cddc3026a9816daa08f38219c4
|
3 |
size 5112
|