Model save
Browse files- README.md +20 -16
- trainer_log.jsonl +19 -0
README.md
CHANGED
@@ -2,10 +2,10 @@
|
|
2 |
license: apache-2.0
|
3 |
library_name: peft
|
4 |
tags:
|
5 |
-
- llama-factory
|
6 |
-
- lora
|
7 |
- trl
|
8 |
- dpo
|
|
|
|
|
9 |
- generated_from_trainer
|
10 |
base_model: mistralai/Mistral-7B-Instruct-v0.2
|
11 |
model-index:
|
@@ -18,19 +18,19 @@ should probably proofread and complete it, then remove this comment. -->
|
|
18 |
|
19 |
# Mistral-7B-Instruct-v0.2-ORPO
|
20 |
|
21 |
-
This model is a fine-tuned version of [mistralai/Mistral-7B-Instruct-v0.2](https://huggingface.co/mistralai/Mistral-7B-Instruct-v0.2) on
|
22 |
It achieves the following results on the evaluation set:
|
23 |
-
- Loss:
|
24 |
-
- Rewards/chosen: -0.
|
25 |
-
- Rewards/rejected: -0.
|
26 |
-
- Rewards/accuracies: 0.
|
27 |
-
- Rewards/margins: 0.
|
28 |
-
- Logps/rejected: -1.
|
29 |
-
- Logps/chosen: -
|
30 |
-
- Logits/rejected: -2.
|
31 |
-
- Logits/chosen: -2.
|
32 |
-
- Sft Loss:
|
33 |
-
- Odds Ratio Loss: 0.
|
34 |
|
35 |
## Model description
|
36 |
|
@@ -51,7 +51,7 @@ More information needed
|
|
51 |
The following hyperparameters were used during training:
|
52 |
- learning_rate: 5e-06
|
53 |
- train_batch_size: 2
|
54 |
-
- eval_batch_size:
|
55 |
- seed: 42
|
56 |
- gradient_accumulation_steps: 8
|
57 |
- total_train_batch_size: 16
|
@@ -59,10 +59,14 @@ The following hyperparameters were used during training:
|
|
59 |
- lr_scheduler_type: cosine
|
60 |
- lr_scheduler_warmup_steps: 0.1
|
61 |
- num_epochs: 3.0
|
62 |
-
- mixed_precision_training: Native AMP
|
63 |
|
64 |
### Training results
|
65 |
|
|
|
|
|
|
|
|
|
|
|
66 |
|
67 |
|
68 |
### Framework versions
|
|
|
2 |
license: apache-2.0
|
3 |
library_name: peft
|
4 |
tags:
|
|
|
|
|
5 |
- trl
|
6 |
- dpo
|
7 |
+
- llama-factory
|
8 |
+
- lora
|
9 |
- generated_from_trainer
|
10 |
base_model: mistralai/Mistral-7B-Instruct-v0.2
|
11 |
model-index:
|
|
|
18 |
|
19 |
# Mistral-7B-Instruct-v0.2-ORPO
|
20 |
|
21 |
+
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.
|
22 |
It achieves the following results on the evaluation set:
|
23 |
+
- Loss: 0.8975
|
24 |
+
- Rewards/chosen: -0.0835
|
25 |
+
- Rewards/rejected: -0.1074
|
26 |
+
- Rewards/accuracies: 0.5900
|
27 |
+
- Rewards/margins: 0.0238
|
28 |
+
- Logps/rejected: -1.0737
|
29 |
+
- Logps/chosen: -0.8352
|
30 |
+
- Logits/rejected: -2.8721
|
31 |
+
- Logits/chosen: -2.8461
|
32 |
+
- Sft Loss: 0.8352
|
33 |
+
- Odds Ratio Loss: 0.6231
|
34 |
|
35 |
## Model description
|
36 |
|
|
|
51 |
The following hyperparameters were used during training:
|
52 |
- learning_rate: 5e-06
|
53 |
- train_batch_size: 2
|
54 |
+
- eval_batch_size: 2
|
55 |
- seed: 42
|
56 |
- gradient_accumulation_steps: 8
|
57 |
- total_train_batch_size: 16
|
|
|
59 |
- lr_scheduler_type: cosine
|
60 |
- lr_scheduler_warmup_steps: 0.1
|
61 |
- num_epochs: 3.0
|
|
|
62 |
|
63 |
### Training results
|
64 |
|
65 |
+
| Training Loss | Epoch | Step | Validation Loss | Rewards/chosen | Rewards/rejected | Rewards/accuracies | Rewards/margins | Logps/rejected | Logps/chosen | Logits/rejected | Logits/chosen | Sft Loss | Odds Ratio Loss |
|
66 |
+
|:-------------:|:------:|:----:|:---------------:|:--------------:|:----------------:|:------------------:|:---------------:|:--------------:|:------------:|:---------------:|:-------------:|:--------:|:---------------:|
|
67 |
+
| 1.0001 | 0.8891 | 500 | 0.9318 | -0.0869 | -0.1112 | 0.5920 | 0.0243 | -1.1123 | -0.8690 | -2.8936 | -2.8713 | 0.8690 | 0.6284 |
|
68 |
+
| 0.906 | 1.7782 | 1000 | 0.9039 | -0.0841 | -0.1081 | 0.5780 | 0.0240 | -1.0811 | -0.8415 | -2.8783 | -2.8533 | 0.8415 | 0.6243 |
|
69 |
+
| 0.9019 | 2.6673 | 1500 | 0.8975 | -0.0835 | -0.1074 | 0.5900 | 0.0238 | -1.0737 | -0.8352 | -2.8721 | -2.8461 | 0.8352 | 0.6231 |
|
70 |
|
71 |
|
72 |
### Framework versions
|
trainer_log.jsonl
CHANGED
@@ -151,3 +151,22 @@
|
|
151 |
{"current_steps": 1490, "total_steps": 1686, "loss": 0.8593, "accuracy": 0.625, "learning_rate": 1.6490167940538343e-07, "epoch": 2.6494776617026004, "percentage": 88.37, "elapsed_time": "4:10:52", "remaining_time": "0:33:00"}
|
152 |
{"current_steps": 1500, "total_steps": 1686, "loss": 0.9019, "accuracy": 0.5874999761581421, "learning_rate": 1.4866882516191339e-07, "epoch": 2.6672593909757722, "percentage": 88.97, "elapsed_time": "4:12:33", "remaining_time": "0:31:18"}
|
153 |
{"current_steps": 1500, "total_steps": 1686, "eval_loss": 0.8975116014480591, "epoch": 2.6672593909757722, "percentage": 88.97, "elapsed_time": "4:15:42", "remaining_time": "0:31:42"}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
151 |
{"current_steps": 1490, "total_steps": 1686, "loss": 0.8593, "accuracy": 0.625, "learning_rate": 1.6490167940538343e-07, "epoch": 2.6494776617026004, "percentage": 88.37, "elapsed_time": "4:10:52", "remaining_time": "0:33:00"}
|
152 |
{"current_steps": 1500, "total_steps": 1686, "loss": 0.9019, "accuracy": 0.5874999761581421, "learning_rate": 1.4866882516191339e-07, "epoch": 2.6672593909757722, "percentage": 88.97, "elapsed_time": "4:12:33", "remaining_time": "0:31:18"}
|
153 |
{"current_steps": 1500, "total_steps": 1686, "eval_loss": 0.8975116014480591, "epoch": 2.6672593909757722, "percentage": 88.97, "elapsed_time": "4:15:42", "remaining_time": "0:31:42"}
|
154 |
+
{"current_steps": 1510, "total_steps": 1686, "loss": 0.8823, "accuracy": 0.5874999761581421, "learning_rate": 1.3325243551706057e-07, "epoch": 2.685041120248944, "percentage": 89.56, "elapsed_time": "4:17:20", "remaining_time": "0:29:59"}
|
155 |
+
{"current_steps": 1520, "total_steps": 1686, "loss": 0.8691, "accuracy": 0.5874999761581421, "learning_rate": 1.1865786358165737e-07, "epoch": 2.702822849522116, "percentage": 90.15, "elapsed_time": "4:18:56", "remaining_time": "0:28:16"}
|
156 |
+
{"current_steps": 1530, "total_steps": 1686, "loss": 0.9517, "accuracy": 0.6187499761581421, "learning_rate": 1.0489017710262311e-07, "epoch": 2.720604578795288, "percentage": 90.75, "elapsed_time": "4:20:35", "remaining_time": "0:26:34"}
|
157 |
+
{"current_steps": 1540, "total_steps": 1686, "loss": 0.8916, "accuracy": 0.5625, "learning_rate": 9.195415670326446e-08, "epoch": 2.73838630806846, "percentage": 91.34, "elapsed_time": "4:22:15", "remaining_time": "0:24:51"}
|
158 |
+
{"current_steps": 1550, "total_steps": 1686, "loss": 0.9067, "accuracy": 0.53125, "learning_rate": 7.985429422327384e-08, "epoch": 2.7561680373416317, "percentage": 91.93, "elapsed_time": "4:23:50", "remaining_time": "0:23:09"}
|
159 |
+
{"current_steps": 1560, "total_steps": 1686, "loss": 0.9293, "accuracy": 0.581250011920929, "learning_rate": 6.859479115900818e-08, "epoch": 2.773949766614803, "percentage": 92.53, "elapsed_time": "4:25:28", "remaining_time": "0:21:26"}
|
160 |
+
{"current_steps": 1570, "total_steps": 1686, "loss": 0.9455, "accuracy": 0.5562499761581421, "learning_rate": 5.817955720457902e-08, "epoch": 2.791731495887975, "percentage": 93.12, "elapsed_time": "4:27:06", "remaining_time": "0:19:44"}
|
161 |
+
{"current_steps": 1580, "total_steps": 1686, "loss": 0.9195, "accuracy": 0.612500011920929, "learning_rate": 4.861220889427199e-08, "epoch": 2.809513225161147, "percentage": 93.71, "elapsed_time": "4:28:40", "remaining_time": "0:18:01"}
|
162 |
+
{"current_steps": 1590, "total_steps": 1686, "loss": 0.9115, "accuracy": 0.5625, "learning_rate": 3.9896068346758074e-08, "epoch": 2.827294954434319, "percentage": 94.31, "elapsed_time": "4:30:16", "remaining_time": "0:16:19"}
|
163 |
+
{"current_steps": 1600, "total_steps": 1686, "loss": 0.9131, "accuracy": 0.5249999761581421, "learning_rate": 3.203416211153832e-08, "epoch": 2.8450766837074903, "percentage": 94.9, "elapsed_time": "4:31:56", "remaining_time": "0:14:36"}
|
164 |
+
{"current_steps": 1610, "total_steps": 1686, "loss": 0.929, "accuracy": 0.5687500238418579, "learning_rate": 2.5029220118019393e-08, "epoch": 2.8628584129806622, "percentage": 95.49, "elapsed_time": "4:33:36", "remaining_time": "0:12:54"}
|
165 |
+
{"current_steps": 1620, "total_steps": 1686, "loss": 0.8816, "accuracy": 0.6312500238418579, "learning_rate": 1.8883674727586122e-08, "epoch": 2.880640142253834, "percentage": 96.09, "elapsed_time": "4:35:12", "remaining_time": "0:11:12"}
|
166 |
+
{"current_steps": 1630, "total_steps": 1686, "loss": 0.8593, "accuracy": 0.5375000238418579, "learning_rate": 1.3599659889000639e-08, "epoch": 2.898421871527006, "percentage": 96.68, "elapsed_time": "4:36:52", "remaining_time": "0:09:30"}
|
167 |
+
{"current_steps": 1640, "total_steps": 1686, "loss": 0.8816, "accuracy": 0.4937500059604645, "learning_rate": 9.179010397421528e-09, "epoch": 2.916203600800178, "percentage": 97.27, "elapsed_time": "4:38:37", "remaining_time": "0:07:48"}
|
168 |
+
{"current_steps": 1650, "total_steps": 1686, "loss": 0.8358, "accuracy": 0.5625, "learning_rate": 5.623261257296509e-09, "epoch": 2.93398533007335, "percentage": 97.86, "elapsed_time": "4:40:14", "remaining_time": "0:06:06"}
|
169 |
+
{"current_steps": 1660, "total_steps": 1686, "loss": 0.8679, "accuracy": 0.543749988079071, "learning_rate": 2.933647149357122e-09, "epoch": 2.9517670593465217, "percentage": 98.46, "elapsed_time": "4:41:52", "remaining_time": "0:04:24"}
|
170 |
+
{"current_steps": 1670, "total_steps": 1686, "loss": 0.9301, "accuracy": 0.53125, "learning_rate": 1.1111020018930717e-09, "epoch": 2.969548788619693, "percentage": 99.05, "elapsed_time": "4:43:31", "remaining_time": "0:02:42"}
|
171 |
+
{"current_steps": 1680, "total_steps": 1686, "loss": 0.8775, "accuracy": 0.6187499761581421, "learning_rate": 1.5625866646051813e-10, "epoch": 2.987330517892865, "percentage": 99.64, "elapsed_time": "4:45:08", "remaining_time": "0:01:01"}
|
172 |
+
{"current_steps": 1686, "total_steps": 1686, "epoch": 2.997999555456768, "percentage": 100.0, "elapsed_time": "4:46:10", "remaining_time": "0:00:00"}
|