Model save
Browse files- README.md +72 -0
- adapter_model.safetensors +1 -1
- all_results.json +9 -0
- train_results.json +9 -0
- trainer_state.json +556 -0
README.md
ADDED
|
@@ -0,0 +1,72 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
license: other
|
| 3 |
+
library_name: peft
|
| 4 |
+
tags:
|
| 5 |
+
- trl
|
| 6 |
+
- dpo
|
| 7 |
+
- generated_from_trainer
|
| 8 |
+
base_model: apple/OpenELM-450M
|
| 9 |
+
model-index:
|
| 10 |
+
- name: OpenELM-450M_lora
|
| 11 |
+
results: []
|
| 12 |
+
---
|
| 13 |
+
|
| 14 |
+
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
|
| 15 |
+
should probably proofread and complete it, then remove this comment. -->
|
| 16 |
+
|
| 17 |
+
# OpenELM-450M_lora
|
| 18 |
+
|
| 19 |
+
This model is a fine-tuned version of [apple/OpenELM-450M](https://huggingface.co/apple/OpenELM-450M) on the None dataset.
|
| 20 |
+
It achieves the following results on the evaluation set:
|
| 21 |
+
- Loss: 0.6940
|
| 22 |
+
- Rewards/chosen: 0.0062
|
| 23 |
+
- Rewards/rejected: 0.0064
|
| 24 |
+
- Rewards/accuracies: 0.4748
|
| 25 |
+
- Rewards/margins: -0.0002
|
| 26 |
+
- Logps/rejected: -567.8893
|
| 27 |
+
- Logps/chosen: -579.9698
|
| 28 |
+
- Logits/rejected: -11.8584
|
| 29 |
+
- Logits/chosen: -12.0367
|
| 30 |
+
|
| 31 |
+
## Model description
|
| 32 |
+
|
| 33 |
+
More information needed
|
| 34 |
+
|
| 35 |
+
## Intended uses & limitations
|
| 36 |
+
|
| 37 |
+
More information needed
|
| 38 |
+
|
| 39 |
+
## Training and evaluation data
|
| 40 |
+
|
| 41 |
+
More information needed
|
| 42 |
+
|
| 43 |
+
## Training procedure
|
| 44 |
+
|
| 45 |
+
### Training hyperparameters
|
| 46 |
+
|
| 47 |
+
The following hyperparameters were used during training:
|
| 48 |
+
- learning_rate: 5e-07
|
| 49 |
+
- train_batch_size: 8
|
| 50 |
+
- eval_batch_size: 8
|
| 51 |
+
- seed: 42
|
| 52 |
+
- gradient_accumulation_steps: 8
|
| 53 |
+
- total_train_batch_size: 64
|
| 54 |
+
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
|
| 55 |
+
- lr_scheduler_type: cosine
|
| 56 |
+
- lr_scheduler_warmup_ratio: 0.1
|
| 57 |
+
- num_epochs: 1
|
| 58 |
+
|
| 59 |
+
### Training results
|
| 60 |
+
|
| 61 |
+
| Training Loss | Epoch | Step | Validation Loss | Rewards/chosen | Rewards/rejected | Rewards/accuracies | Rewards/margins | Logps/rejected | Logps/chosen | Logits/rejected | Logits/chosen |
|
| 62 |
+
|:-------------:|:------:|:----:|:---------------:|:--------------:|:----------------:|:------------------:|:---------------:|:--------------:|:------------:|:---------------:|:-------------:|
|
| 63 |
+
| 0.6943 | 0.8975 | 300 | 0.6940 | 0.0062 | 0.0064 | 0.4748 | -0.0002 | -567.8893 | -579.9698 | -11.8584 | -12.0367 |
|
| 64 |
+
|
| 65 |
+
|
| 66 |
+
### Framework versions
|
| 67 |
+
|
| 68 |
+
- PEFT 0.11.0
|
| 69 |
+
- Transformers 4.40.2
|
| 70 |
+
- Pytorch 2.3.0+cu121
|
| 71 |
+
- Datasets 2.19.1
|
| 72 |
+
- Tokenizers 0.19.1
|
adapter_model.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 7670448
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:ac4dd479051fdc10ffeda881ae3924cd935a5ac91f0d54c722462201c441f21a
|
| 3 |
size 7670448
|
all_results.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 0.9992520568436799,
|
| 3 |
+
"total_flos": 0.0,
|
| 4 |
+
"train_loss": 0.6938572500994106,
|
| 5 |
+
"train_runtime": 20825.8975,
|
| 6 |
+
"train_samples": 21390,
|
| 7 |
+
"train_samples_per_second": 1.027,
|
| 8 |
+
"train_steps_per_second": 0.016
|
| 9 |
+
}
|
train_results.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 0.9992520568436799,
|
| 3 |
+
"total_flos": 0.0,
|
| 4 |
+
"train_loss": 0.6938572500994106,
|
| 5 |
+
"train_runtime": 20825.8975,
|
| 6 |
+
"train_samples": 21390,
|
| 7 |
+
"train_samples_per_second": 1.027,
|
| 8 |
+
"train_steps_per_second": 0.016
|
| 9 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,556 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 0.9992520568436799,
|
| 5 |
+
"eval_steps": 300,
|
| 6 |
+
"global_step": 334,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.0029917726252804786,
|
| 13 |
+
"grad_norm": 0.515625,
|
| 14 |
+
"learning_rate": 1.4705882352941176e-08,
|
| 15 |
+
"logits/chosen": -11.9029541015625,
|
| 16 |
+
"logits/rejected": -11.867537498474121,
|
| 17 |
+
"logps/chosen": -446.77239990234375,
|
| 18 |
+
"logps/rejected": -451.92742919921875,
|
| 19 |
+
"loss": 0.6931,
|
| 20 |
+
"rewards/accuracies": 0.0,
|
| 21 |
+
"rewards/chosen": 0.0,
|
| 22 |
+
"rewards/margins": 0.0,
|
| 23 |
+
"rewards/rejected": 0.0,
|
| 24 |
+
"step": 1
|
| 25 |
+
},
|
| 26 |
+
{
|
| 27 |
+
"epoch": 0.029917726252804786,
|
| 28 |
+
"grad_norm": 2.84375,
|
| 29 |
+
"learning_rate": 1.4705882352941175e-07,
|
| 30 |
+
"logits/chosen": -11.970257759094238,
|
| 31 |
+
"logits/rejected": -11.80525016784668,
|
| 32 |
+
"logps/chosen": -548.0371704101562,
|
| 33 |
+
"logps/rejected": -528.268798828125,
|
| 34 |
+
"loss": 0.693,
|
| 35 |
+
"rewards/accuracies": 0.4184027910232544,
|
| 36 |
+
"rewards/chosen": 0.0011305785737931728,
|
| 37 |
+
"rewards/margins": 0.0008499751565977931,
|
| 38 |
+
"rewards/rejected": 0.0002806035045068711,
|
| 39 |
+
"step": 10
|
| 40 |
+
},
|
| 41 |
+
{
|
| 42 |
+
"epoch": 0.05983545250560957,
|
| 43 |
+
"grad_norm": 1.484375,
|
| 44 |
+
"learning_rate": 2.941176470588235e-07,
|
| 45 |
+
"logits/chosen": -12.091507911682129,
|
| 46 |
+
"logits/rejected": -11.931904792785645,
|
| 47 |
+
"logps/chosen": -605.2603149414062,
|
| 48 |
+
"logps/rejected": -594.6846923828125,
|
| 49 |
+
"loss": 0.6946,
|
| 50 |
+
"rewards/accuracies": 0.4781250059604645,
|
| 51 |
+
"rewards/chosen": -0.0016426166985183954,
|
| 52 |
+
"rewards/margins": -0.001760849030688405,
|
| 53 |
+
"rewards/rejected": 0.00011823121167253703,
|
| 54 |
+
"step": 20
|
| 55 |
+
},
|
| 56 |
+
{
|
| 57 |
+
"epoch": 0.08975317875841436,
|
| 58 |
+
"grad_norm": 0.8125,
|
| 59 |
+
"learning_rate": 4.4117647058823526e-07,
|
| 60 |
+
"logits/chosen": -12.251832962036133,
|
| 61 |
+
"logits/rejected": -11.93907356262207,
|
| 62 |
+
"logps/chosen": -627.1741333007812,
|
| 63 |
+
"logps/rejected": -592.0576782226562,
|
| 64 |
+
"loss": 0.6927,
|
| 65 |
+
"rewards/accuracies": 0.4765625,
|
| 66 |
+
"rewards/chosen": -0.013599636033177376,
|
| 67 |
+
"rewards/margins": 0.001697429222986102,
|
| 68 |
+
"rewards/rejected": -0.015297065488994122,
|
| 69 |
+
"step": 30
|
| 70 |
+
},
|
| 71 |
+
{
|
| 72 |
+
"epoch": 0.11967090501121914,
|
| 73 |
+
"grad_norm": 0.7734375,
|
| 74 |
+
"learning_rate": 4.995066821070679e-07,
|
| 75 |
+
"logits/chosen": -12.038887023925781,
|
| 76 |
+
"logits/rejected": -11.891900062561035,
|
| 77 |
+
"logps/chosen": -573.0090942382812,
|
| 78 |
+
"logps/rejected": -555.4964599609375,
|
| 79 |
+
"loss": 0.6966,
|
| 80 |
+
"rewards/accuracies": 0.4390625059604645,
|
| 81 |
+
"rewards/chosen": -0.009484687820076942,
|
| 82 |
+
"rewards/margins": -0.005814659409224987,
|
| 83 |
+
"rewards/rejected": -0.0036700288765132427,
|
| 84 |
+
"step": 40
|
| 85 |
+
},
|
| 86 |
+
{
|
| 87 |
+
"epoch": 0.14958863126402394,
|
| 88 |
+
"grad_norm": 1.1796875,
|
| 89 |
+
"learning_rate": 4.964990092676262e-07,
|
| 90 |
+
"logits/chosen": -12.046560287475586,
|
| 91 |
+
"logits/rejected": -11.745036125183105,
|
| 92 |
+
"logps/chosen": -487.1591796875,
|
| 93 |
+
"logps/rejected": -467.3463439941406,
|
| 94 |
+
"loss": 0.6961,
|
| 95 |
+
"rewards/accuracies": 0.4625000059604645,
|
| 96 |
+
"rewards/chosen": -0.014206953346729279,
|
| 97 |
+
"rewards/margins": -0.0039894962683320045,
|
| 98 |
+
"rewards/rejected": -0.010217458009719849,
|
| 99 |
+
"step": 50
|
| 100 |
+
},
|
| 101 |
+
{
|
| 102 |
+
"epoch": 0.17950635751682872,
|
| 103 |
+
"grad_norm": 0.65234375,
|
| 104 |
+
"learning_rate": 4.907906416994145e-07,
|
| 105 |
+
"logits/chosen": -11.951323509216309,
|
| 106 |
+
"logits/rejected": -11.81787109375,
|
| 107 |
+
"logps/chosen": -524.3377685546875,
|
| 108 |
+
"logps/rejected": -518.5364990234375,
|
| 109 |
+
"loss": 0.6951,
|
| 110 |
+
"rewards/accuracies": 0.4906249940395355,
|
| 111 |
+
"rewards/chosen": -0.007045179605484009,
|
| 112 |
+
"rewards/margins": -0.0024539525620639324,
|
| 113 |
+
"rewards/rejected": -0.004591226577758789,
|
| 114 |
+
"step": 60
|
| 115 |
+
},
|
| 116 |
+
{
|
| 117 |
+
"epoch": 0.2094240837696335,
|
| 118 |
+
"grad_norm": 0.7109375,
|
| 119 |
+
"learning_rate": 4.824441214720628e-07,
|
| 120 |
+
"logits/chosen": -12.155089378356934,
|
| 121 |
+
"logits/rejected": -11.935697555541992,
|
| 122 |
+
"logps/chosen": -614.5479736328125,
|
| 123 |
+
"logps/rejected": -598.7986450195312,
|
| 124 |
+
"loss": 0.694,
|
| 125 |
+
"rewards/accuracies": 0.4921875,
|
| 126 |
+
"rewards/chosen": -0.004119172692298889,
|
| 127 |
+
"rewards/margins": -0.0007487249677069485,
|
| 128 |
+
"rewards/rejected": -0.003370448248460889,
|
| 129 |
+
"step": 70
|
| 130 |
+
},
|
| 131 |
+
{
|
| 132 |
+
"epoch": 0.2393418100224383,
|
| 133 |
+
"grad_norm": 1.125,
|
| 134 |
+
"learning_rate": 4.7155089480780365e-07,
|
| 135 |
+
"logits/chosen": -12.310976028442383,
|
| 136 |
+
"logits/rejected": -11.926130294799805,
|
| 137 |
+
"logps/chosen": -590.0772705078125,
|
| 138 |
+
"logps/rejected": -552.1112060546875,
|
| 139 |
+
"loss": 0.694,
|
| 140 |
+
"rewards/accuracies": 0.4703125059604645,
|
| 141 |
+
"rewards/chosen": -0.014592505991458893,
|
| 142 |
+
"rewards/margins": -0.000543795176781714,
|
| 143 |
+
"rewards/rejected": -0.014048713259398937,
|
| 144 |
+
"step": 80
|
| 145 |
+
},
|
| 146 |
+
{
|
| 147 |
+
"epoch": 0.26925953627524307,
|
| 148 |
+
"grad_norm": 1.7890625,
|
| 149 |
+
"learning_rate": 4.582303101775248e-07,
|
| 150 |
+
"logits/chosen": -12.02568244934082,
|
| 151 |
+
"logits/rejected": -11.863547325134277,
|
| 152 |
+
"logps/chosen": -596.1324462890625,
|
| 153 |
+
"logps/rejected": -583.9989624023438,
|
| 154 |
+
"loss": 0.6946,
|
| 155 |
+
"rewards/accuracies": 0.4625000059604645,
|
| 156 |
+
"rewards/chosen": -0.00669059157371521,
|
| 157 |
+
"rewards/margins": -0.0019042941275984049,
|
| 158 |
+
"rewards/rejected": -0.004786298610270023,
|
| 159 |
+
"step": 90
|
| 160 |
+
},
|
| 161 |
+
{
|
| 162 |
+
"epoch": 0.2991772625280479,
|
| 163 |
+
"grad_norm": 1.1015625,
|
| 164 |
+
"learning_rate": 4.426283106939473e-07,
|
| 165 |
+
"logits/chosen": -12.22739028930664,
|
| 166 |
+
"logits/rejected": -11.815451622009277,
|
| 167 |
+
"logps/chosen": -604.3762817382812,
|
| 168 |
+
"logps/rejected": -574.3903198242188,
|
| 169 |
+
"loss": 0.6949,
|
| 170 |
+
"rewards/accuracies": 0.4828124940395355,
|
| 171 |
+
"rewards/chosen": 0.0030080166179686785,
|
| 172 |
+
"rewards/margins": -0.0021901517175137997,
|
| 173 |
+
"rewards/rejected": 0.0051981681026518345,
|
| 174 |
+
"step": 100
|
| 175 |
+
},
|
| 176 |
+
{
|
| 177 |
+
"epoch": 0.32909498878085264,
|
| 178 |
+
"grad_norm": 1.453125,
|
| 179 |
+
"learning_rate": 4.249158351283413e-07,
|
| 180 |
+
"logits/chosen": -12.154890060424805,
|
| 181 |
+
"logits/rejected": -11.859006881713867,
|
| 182 |
+
"logps/chosen": -524.4495849609375,
|
| 183 |
+
"logps/rejected": -509.00946044921875,
|
| 184 |
+
"loss": 0.6954,
|
| 185 |
+
"rewards/accuracies": 0.48124998807907104,
|
| 186 |
+
"rewards/chosen": 0.002577757928520441,
|
| 187 |
+
"rewards/margins": -0.002684831153601408,
|
| 188 |
+
"rewards/rejected": 0.005262589547783136,
|
| 189 |
+
"step": 110
|
| 190 |
+
},
|
| 191 |
+
{
|
| 192 |
+
"epoch": 0.35901271503365745,
|
| 193 |
+
"grad_norm": 3.0625,
|
| 194 |
+
"learning_rate": 4.0528694506957754e-07,
|
| 195 |
+
"logits/chosen": -12.192368507385254,
|
| 196 |
+
"logits/rejected": -12.028105735778809,
|
| 197 |
+
"logps/chosen": -528.8504028320312,
|
| 198 |
+
"logps/rejected": -517.0180053710938,
|
| 199 |
+
"loss": 0.693,
|
| 200 |
+
"rewards/accuracies": 0.5015624761581421,
|
| 201 |
+
"rewards/chosen": 0.007784114684909582,
|
| 202 |
+
"rewards/margins": 0.0019344612956047058,
|
| 203 |
+
"rewards/rejected": 0.005849653389304876,
|
| 204 |
+
"step": 120
|
| 205 |
+
},
|
| 206 |
+
{
|
| 207 |
+
"epoch": 0.3889304412864622,
|
| 208 |
+
"grad_norm": 1.0546875,
|
| 209 |
+
"learning_rate": 3.839566987447491e-07,
|
| 210 |
+
"logits/chosen": -12.14574146270752,
|
| 211 |
+
"logits/rejected": -12.06743335723877,
|
| 212 |
+
"logps/chosen": -646.7266845703125,
|
| 213 |
+
"logps/rejected": -643.1847534179688,
|
| 214 |
+
"loss": 0.6923,
|
| 215 |
+
"rewards/accuracies": 0.4906249940395355,
|
| 216 |
+
"rewards/chosen": -0.00342937046661973,
|
| 217 |
+
"rewards/margins": 0.0035477778874337673,
|
| 218 |
+
"rewards/rejected": -0.006977148354053497,
|
| 219 |
+
"step": 130
|
| 220 |
+
},
|
| 221 |
+
{
|
| 222 |
+
"epoch": 0.418848167539267,
|
| 223 |
+
"grad_norm": 0.7421875,
|
| 224 |
+
"learning_rate": 3.6115879479623183e-07,
|
| 225 |
+
"logits/chosen": -11.896639823913574,
|
| 226 |
+
"logits/rejected": -11.759775161743164,
|
| 227 |
+
"logps/chosen": -548.5411376953125,
|
| 228 |
+
"logps/rejected": -533.6222534179688,
|
| 229 |
+
"loss": 0.6943,
|
| 230 |
+
"rewards/accuracies": 0.4984374940395355,
|
| 231 |
+
"rewards/chosen": -0.006624075584113598,
|
| 232 |
+
"rewards/margins": -0.00133815489243716,
|
| 233 |
+
"rewards/rejected": -0.005285919643938541,
|
| 234 |
+
"step": 140
|
| 235 |
+
},
|
| 236 |
+
{
|
| 237 |
+
"epoch": 0.4487658937920718,
|
| 238 |
+
"grad_norm": 0.87109375,
|
| 239 |
+
"learning_rate": 3.371430118304538e-07,
|
| 240 |
+
"logits/chosen": -11.987521171569824,
|
| 241 |
+
"logits/rejected": -11.82154655456543,
|
| 242 |
+
"logps/chosen": -599.1915283203125,
|
| 243 |
+
"logps/rejected": -586.5440673828125,
|
| 244 |
+
"loss": 0.6945,
|
| 245 |
+
"rewards/accuracies": 0.4593749940395355,
|
| 246 |
+
"rewards/chosen": -0.0005397430504672229,
|
| 247 |
+
"rewards/margins": -0.001651047496125102,
|
| 248 |
+
"rewards/rejected": 0.0011113042710348964,
|
| 249 |
+
"step": 150
|
| 250 |
+
},
|
| 251 |
+
{
|
| 252 |
+
"epoch": 0.4786836200448766,
|
| 253 |
+
"grad_norm": 2.078125,
|
| 254 |
+
"learning_rate": 3.121724717912138e-07,
|
| 255 |
+
"logits/chosen": -12.010518074035645,
|
| 256 |
+
"logits/rejected": -11.801678657531738,
|
| 257 |
+
"logps/chosen": -477.989013671875,
|
| 258 |
+
"logps/rejected": -461.06695556640625,
|
| 259 |
+
"loss": 0.6923,
|
| 260 |
+
"rewards/accuracies": 0.5,
|
| 261 |
+
"rewards/chosen": -0.002107062842696905,
|
| 262 |
+
"rewards/margins": 0.002797911176458001,
|
| 263 |
+
"rewards/rejected": -0.00490497425198555,
|
| 264 |
+
"step": 160
|
| 265 |
+
},
|
| 266 |
+
{
|
| 267 |
+
"epoch": 0.5086013462976814,
|
| 268 |
+
"grad_norm": 1.046875,
|
| 269 |
+
"learning_rate": 2.865207571406029e-07,
|
| 270 |
+
"logits/chosen": -12.087748527526855,
|
| 271 |
+
"logits/rejected": -11.823697090148926,
|
| 272 |
+
"logps/chosen": -565.2094116210938,
|
| 273 |
+
"logps/rejected": -535.0115966796875,
|
| 274 |
+
"loss": 0.6945,
|
| 275 |
+
"rewards/accuracies": 0.45625001192092896,
|
| 276 |
+
"rewards/chosen": -0.010714416392147541,
|
| 277 |
+
"rewards/margins": -0.0014880959643051028,
|
| 278 |
+
"rewards/rejected": -0.009226320311427116,
|
| 279 |
+
"step": 170
|
| 280 |
+
},
|
| 281 |
+
{
|
| 282 |
+
"epoch": 0.5385190725504861,
|
| 283 |
+
"grad_norm": 0.69140625,
|
| 284 |
+
"learning_rate": 2.6046891343229986e-07,
|
| 285 |
+
"logits/chosen": -12.05897331237793,
|
| 286 |
+
"logits/rejected": -11.826770782470703,
|
| 287 |
+
"logps/chosen": -639.4141845703125,
|
| 288 |
+
"logps/rejected": -625.544189453125,
|
| 289 |
+
"loss": 0.694,
|
| 290 |
+
"rewards/accuracies": 0.4703125059604645,
|
| 291 |
+
"rewards/chosen": -0.008839543908834457,
|
| 292 |
+
"rewards/margins": -0.0007707075565122068,
|
| 293 |
+
"rewards/rejected": -0.008068837225437164,
|
| 294 |
+
"step": 180
|
| 295 |
+
},
|
| 296 |
+
{
|
| 297 |
+
"epoch": 0.5684367988032909,
|
| 298 |
+
"grad_norm": 2.15625,
|
| 299 |
+
"learning_rate": 2.3430237011767164e-07,
|
| 300 |
+
"logits/chosen": -12.140950202941895,
|
| 301 |
+
"logits/rejected": -11.830968856811523,
|
| 302 |
+
"logps/chosen": -564.7281494140625,
|
| 303 |
+
"logps/rejected": -540.7762451171875,
|
| 304 |
+
"loss": 0.6931,
|
| 305 |
+
"rewards/accuracies": 0.484375,
|
| 306 |
+
"rewards/chosen": -0.00785023346543312,
|
| 307 |
+
"rewards/margins": 0.0012117780279368162,
|
| 308 |
+
"rewards/rejected": -0.00906201172620058,
|
| 309 |
+
"step": 190
|
| 310 |
+
},
|
| 311 |
+
{
|
| 312 |
+
"epoch": 0.5983545250560958,
|
| 313 |
+
"grad_norm": 0.64453125,
|
| 314 |
+
"learning_rate": 2.0830781332097445e-07,
|
| 315 |
+
"logits/chosen": -12.064250946044922,
|
| 316 |
+
"logits/rejected": -11.847702026367188,
|
| 317 |
+
"logps/chosen": -601.1182861328125,
|
| 318 |
+
"logps/rejected": -584.1577758789062,
|
| 319 |
+
"loss": 0.6932,
|
| 320 |
+
"rewards/accuracies": 0.46562498807907104,
|
| 321 |
+
"rewards/chosen": -0.0038043882232159376,
|
| 322 |
+
"rewards/margins": 0.0006765492144040763,
|
| 323 |
+
"rewards/rejected": -0.0044809365645051,
|
| 324 |
+
"step": 200
|
| 325 |
+
},
|
| 326 |
+
{
|
| 327 |
+
"epoch": 0.6282722513089005,
|
| 328 |
+
"grad_norm": 0.6796875,
|
| 329 |
+
"learning_rate": 1.8277004484618357e-07,
|
| 330 |
+
"logits/chosen": -12.159658432006836,
|
| 331 |
+
"logits/rejected": -11.93867015838623,
|
| 332 |
+
"logps/chosen": -570.8473510742188,
|
| 333 |
+
"logps/rejected": -558.4066162109375,
|
| 334 |
+
"loss": 0.6956,
|
| 335 |
+
"rewards/accuracies": 0.46562498807907104,
|
| 336 |
+
"rewards/chosen": -0.01011698879301548,
|
| 337 |
+
"rewards/margins": -0.003837681608274579,
|
| 338 |
+
"rewards/rejected": -0.006279306020587683,
|
| 339 |
+
"step": 210
|
| 340 |
+
},
|
| 341 |
+
{
|
| 342 |
+
"epoch": 0.6581899775617053,
|
| 343 |
+
"grad_norm": 0.90625,
|
| 344 |
+
"learning_rate": 1.579688618288305e-07,
|
| 345 |
+
"logits/chosen": -12.136404991149902,
|
| 346 |
+
"logits/rejected": -11.933469772338867,
|
| 347 |
+
"logps/chosen": -653.62841796875,
|
| 348 |
+
"logps/rejected": -629.8015747070312,
|
| 349 |
+
"loss": 0.6917,
|
| 350 |
+
"rewards/accuracies": 0.5062500238418579,
|
| 351 |
+
"rewards/chosen": 0.0027369544841349125,
|
| 352 |
+
"rewards/margins": 0.003812599228695035,
|
| 353 |
+
"rewards/rejected": -0.0010756452102214098,
|
| 354 |
+
"step": 220
|
| 355 |
+
},
|
| 356 |
+
{
|
| 357 |
+
"epoch": 0.6881077038145101,
|
| 358 |
+
"grad_norm": 1.046875,
|
| 359 |
+
"learning_rate": 1.341759912200346e-07,
|
| 360 |
+
"logits/chosen": -12.061933517456055,
|
| 361 |
+
"logits/rejected": -11.955540657043457,
|
| 362 |
+
"logps/chosen": -541.6209716796875,
|
| 363 |
+
"logps/rejected": -530.1668701171875,
|
| 364 |
+
"loss": 0.69,
|
| 365 |
+
"rewards/accuracies": 0.526562511920929,
|
| 366 |
+
"rewards/chosen": 0.006943908985704184,
|
| 367 |
+
"rewards/margins": 0.007641922682523727,
|
| 368 |
+
"rewards/rejected": -0.0006980125908739865,
|
| 369 |
+
"step": 230
|
| 370 |
+
},
|
| 371 |
+
{
|
| 372 |
+
"epoch": 0.7180254300673149,
|
| 373 |
+
"grad_norm": 2.078125,
|
| 374 |
+
"learning_rate": 1.11652112689164e-07,
|
| 375 |
+
"logits/chosen": -11.996115684509277,
|
| 376 |
+
"logits/rejected": -11.906721115112305,
|
| 377 |
+
"logps/chosen": -514.420166015625,
|
| 378 |
+
"logps/rejected": -510.46307373046875,
|
| 379 |
+
"loss": 0.6932,
|
| 380 |
+
"rewards/accuracies": 0.5062500238418579,
|
| 381 |
+
"rewards/chosen": 0.004082207567989826,
|
| 382 |
+
"rewards/margins": 0.0010683867149055004,
|
| 383 |
+
"rewards/rejected": 0.003013820853084326,
|
| 384 |
+
"step": 240
|
| 385 |
+
},
|
| 386 |
+
{
|
| 387 |
+
"epoch": 0.7479431563201197,
|
| 388 |
+
"grad_norm": 0.60546875,
|
| 389 |
+
"learning_rate": 9.064400256282755e-08,
|
| 390 |
+
"logits/chosen": -11.868894577026367,
|
| 391 |
+
"logits/rejected": -11.739983558654785,
|
| 392 |
+
"logps/chosen": -636.687744140625,
|
| 393 |
+
"logps/rejected": -622.6748657226562,
|
| 394 |
+
"loss": 0.6937,
|
| 395 |
+
"rewards/accuracies": 0.4984374940395355,
|
| 396 |
+
"rewards/chosen": 0.007722427137196064,
|
| 397 |
+
"rewards/margins": -8.60728334828309e-07,
|
| 398 |
+
"rewards/rejected": 0.007723286747932434,
|
| 399 |
+
"step": 250
|
| 400 |
+
},
|
| 401 |
+
{
|
| 402 |
+
"epoch": 0.7778608825729244,
|
| 403 |
+
"grad_norm": 0.625,
|
| 404 |
+
"learning_rate": 7.138183009179921e-08,
|
| 405 |
+
"logits/chosen": -11.95314884185791,
|
| 406 |
+
"logits/rejected": -11.800286293029785,
|
| 407 |
+
"logps/chosen": -516.5999755859375,
|
| 408 |
+
"logps/rejected": -502.53424072265625,
|
| 409 |
+
"loss": 0.6941,
|
| 410 |
+
"rewards/accuracies": 0.46875,
|
| 411 |
+
"rewards/chosen": 0.003324865596368909,
|
| 412 |
+
"rewards/margins": -0.0007609135354869068,
|
| 413 |
+
"rewards/rejected": 0.004085779655724764,
|
| 414 |
+
"step": 260
|
| 415 |
+
},
|
| 416 |
+
{
|
| 417 |
+
"epoch": 0.8077786088257293,
|
| 418 |
+
"grad_norm": 2.03125,
|
| 419 |
+
"learning_rate": 5.4076635668540065e-08,
|
| 420 |
+
"logits/chosen": -11.870991706848145,
|
| 421 |
+
"logits/rejected": -11.772178649902344,
|
| 422 |
+
"logps/chosen": -537.8231201171875,
|
| 423 |
+
"logps/rejected": -536.79296875,
|
| 424 |
+
"loss": 0.6933,
|
| 425 |
+
"rewards/accuracies": 0.48124998807907104,
|
| 426 |
+
"rewards/chosen": -0.0003947736695408821,
|
| 427 |
+
"rewards/margins": 0.0007606123690493405,
|
| 428 |
+
"rewards/rejected": -0.0011553869117051363,
|
| 429 |
+
"step": 270
|
| 430 |
+
},
|
| 431 |
+
{
|
| 432 |
+
"epoch": 0.837696335078534,
|
| 433 |
+
"grad_norm": 0.59765625,
|
| 434 |
+
"learning_rate": 3.8918018624496286e-08,
|
| 435 |
+
"logits/chosen": -12.036161422729492,
|
| 436 |
+
"logits/rejected": -11.860921859741211,
|
| 437 |
+
"logps/chosen": -554.2711791992188,
|
| 438 |
+
"logps/rejected": -541.3438720703125,
|
| 439 |
+
"loss": 0.6953,
|
| 440 |
+
"rewards/accuracies": 0.4781250059604645,
|
| 441 |
+
"rewards/chosen": 0.00031470804242417216,
|
| 442 |
+
"rewards/margins": -0.0031545311212539673,
|
| 443 |
+
"rewards/rejected": 0.0034692403860390186,
|
| 444 |
+
"step": 280
|
| 445 |
+
},
|
| 446 |
+
{
|
| 447 |
+
"epoch": 0.8676140613313388,
|
| 448 |
+
"grad_norm": 2.578125,
|
| 449 |
+
"learning_rate": 2.6072059940146772e-08,
|
| 450 |
+
"logits/chosen": -12.018091201782227,
|
| 451 |
+
"logits/rejected": -11.894245147705078,
|
| 452 |
+
"logps/chosen": -569.8294677734375,
|
| 453 |
+
"logps/rejected": -563.2034912109375,
|
| 454 |
+
"loss": 0.6936,
|
| 455 |
+
"rewards/accuracies": 0.4921875,
|
| 456 |
+
"rewards/chosen": 0.004260566085577011,
|
| 457 |
+
"rewards/margins": 0.0009015941177494824,
|
| 458 |
+
"rewards/rejected": 0.003358972491696477,
|
| 459 |
+
"step": 290
|
| 460 |
+
},
|
| 461 |
+
{
|
| 462 |
+
"epoch": 0.8975317875841436,
|
| 463 |
+
"grad_norm": 1.015625,
|
| 464 |
+
"learning_rate": 1.5679502627027136e-08,
|
| 465 |
+
"logits/chosen": -12.114303588867188,
|
| 466 |
+
"logits/rejected": -11.98996639251709,
|
| 467 |
+
"logps/chosen": -567.474853515625,
|
| 468 |
+
"logps/rejected": -552.4931640625,
|
| 469 |
+
"loss": 0.6943,
|
| 470 |
+
"rewards/accuracies": 0.46406251192092896,
|
| 471 |
+
"rewards/chosen": 0.0020927421282976866,
|
| 472 |
+
"rewards/margins": -0.0002106330794049427,
|
| 473 |
+
"rewards/rejected": 0.0023033744655549526,
|
| 474 |
+
"step": 300
|
| 475 |
+
},
|
| 476 |
+
{
|
| 477 |
+
"epoch": 0.8975317875841436,
|
| 478 |
+
"eval_logits/chosen": -12.036696434020996,
|
| 479 |
+
"eval_logits/rejected": -11.858353614807129,
|
| 480 |
+
"eval_logps/chosen": -579.9698486328125,
|
| 481 |
+
"eval_logps/rejected": -567.8892822265625,
|
| 482 |
+
"eval_loss": 0.6939929127693176,
|
| 483 |
+
"eval_rewards/accuracies": 0.4747757911682129,
|
| 484 |
+
"eval_rewards/chosen": 0.0061609940603375435,
|
| 485 |
+
"eval_rewards/margins": -0.00024100362497847527,
|
| 486 |
+
"eval_rewards/rejected": 0.006401997059583664,
|
| 487 |
+
"eval_runtime": 2311.0773,
|
| 488 |
+
"eval_samples_per_second": 2.314,
|
| 489 |
+
"eval_steps_per_second": 0.289,
|
| 490 |
+
"step": 300
|
| 491 |
+
},
|
| 492 |
+
{
|
| 493 |
+
"epoch": 0.9274495138369484,
|
| 494 |
+
"grad_norm": 0.7890625,
|
| 495 |
+
"learning_rate": 7.85420971784223e-09,
|
| 496 |
+
"logits/chosen": -12.027294158935547,
|
| 497 |
+
"logits/rejected": -11.775075912475586,
|
| 498 |
+
"logps/chosen": -637.2244262695312,
|
| 499 |
+
"logps/rejected": -614.4982299804688,
|
| 500 |
+
"loss": 0.6929,
|
| 501 |
+
"rewards/accuracies": 0.49687498807907104,
|
| 502 |
+
"rewards/chosen": 0.008133328519761562,
|
| 503 |
+
"rewards/margins": 0.0024830640759319067,
|
| 504 |
+
"rewards/rejected": 0.005650263279676437,
|
| 505 |
+
"step": 310
|
| 506 |
+
},
|
| 507 |
+
{
|
| 508 |
+
"epoch": 0.9573672400897532,
|
| 509 |
+
"grad_norm": 0.99609375,
|
| 510 |
+
"learning_rate": 2.6819167592529168e-09,
|
| 511 |
+
"logits/chosen": -11.920928955078125,
|
| 512 |
+
"logits/rejected": -11.692464828491211,
|
| 513 |
+
"logps/chosen": -622.5579223632812,
|
| 514 |
+
"logps/rejected": -609.9359130859375,
|
| 515 |
+
"loss": 0.693,
|
| 516 |
+
"rewards/accuracies": 0.48906248807907104,
|
| 517 |
+
"rewards/chosen": 0.012452816590666771,
|
| 518 |
+
"rewards/margins": 0.0013520055217668414,
|
| 519 |
+
"rewards/rejected": 0.011100810021162033,
|
| 520 |
+
"step": 320
|
| 521 |
+
},
|
| 522 |
+
{
|
| 523 |
+
"epoch": 0.9872849663425579,
|
| 524 |
+
"grad_norm": 1.734375,
|
| 525 |
+
"learning_rate": 2.1929247528540418e-10,
|
| 526 |
+
"logits/chosen": -12.089780807495117,
|
| 527 |
+
"logits/rejected": -11.94688606262207,
|
| 528 |
+
"logps/chosen": -568.0908203125,
|
| 529 |
+
"logps/rejected": -560.9391479492188,
|
| 530 |
+
"loss": 0.695,
|
| 531 |
+
"rewards/accuracies": 0.512499988079071,
|
| 532 |
+
"rewards/chosen": 0.013248354196548462,
|
| 533 |
+
"rewards/margins": -0.0018901375588029623,
|
| 534 |
+
"rewards/rejected": 0.015138491988182068,
|
| 535 |
+
"step": 330
|
| 536 |
+
},
|
| 537 |
+
{
|
| 538 |
+
"epoch": 0.9992520568436799,
|
| 539 |
+
"step": 334,
|
| 540 |
+
"total_flos": 0.0,
|
| 541 |
+
"train_loss": 0.6938572500994106,
|
| 542 |
+
"train_runtime": 20825.8975,
|
| 543 |
+
"train_samples_per_second": 1.027,
|
| 544 |
+
"train_steps_per_second": 0.016
|
| 545 |
+
}
|
| 546 |
+
],
|
| 547 |
+
"logging_steps": 10,
|
| 548 |
+
"max_steps": 334,
|
| 549 |
+
"num_input_tokens_seen": 0,
|
| 550 |
+
"num_train_epochs": 1,
|
| 551 |
+
"save_steps": 300,
|
| 552 |
+
"total_flos": 0.0,
|
| 553 |
+
"train_batch_size": 8,
|
| 554 |
+
"trial_name": null,
|
| 555 |
+
"trial_params": null
|
| 556 |
+
}
|