Model save
Browse files- README.md +60 -0
- all_results.json +9 -0
- generation_config.json +12 -0
- train_results.json +9 -0
- trainer_state.json +432 -0
README.md
ADDED
|
@@ -0,0 +1,60 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
license: llama3
|
| 3 |
+
base_model: Jackie999/llama3-sudo-5epochs-tofu_full_sft
|
| 4 |
+
tags:
|
| 5 |
+
- trl
|
| 6 |
+
- dpo
|
| 7 |
+
- generated_from_trainer
|
| 8 |
+
model-index:
|
| 9 |
+
- name: llama3-sudo-dpo-10epochs-forget10mix400-1sft-2fullpara-1e-5
|
| 10 |
+
results: []
|
| 11 |
+
---
|
| 12 |
+
|
| 13 |
+
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
|
| 14 |
+
should probably proofread and complete it, then remove this comment. -->
|
| 15 |
+
|
| 16 |
+
# llama3-sudo-dpo-10epochs-forget10mix400-1sft-2fullpara-1e-5
|
| 17 |
+
|
| 18 |
+
This model is a fine-tuned version of [Jackie999/llama3-sudo-5epochs-tofu_full_sft](https://huggingface.co/Jackie999/llama3-sudo-5epochs-tofu_full_sft) on an unknown dataset.
|
| 19 |
+
|
| 20 |
+
## Model description
|
| 21 |
+
|
| 22 |
+
More information needed
|
| 23 |
+
|
| 24 |
+
## Intended uses & limitations
|
| 25 |
+
|
| 26 |
+
More information needed
|
| 27 |
+
|
| 28 |
+
## Training and evaluation data
|
| 29 |
+
|
| 30 |
+
More information needed
|
| 31 |
+
|
| 32 |
+
## Training procedure
|
| 33 |
+
|
| 34 |
+
### Training hyperparameters
|
| 35 |
+
|
| 36 |
+
The following hyperparameters were used during training:
|
| 37 |
+
- learning_rate: 1e-05
|
| 38 |
+
- train_batch_size: 4
|
| 39 |
+
- eval_batch_size: 8
|
| 40 |
+
- seed: 42
|
| 41 |
+
- distributed_type: multi-GPU
|
| 42 |
+
- num_devices: 4
|
| 43 |
+
- gradient_accumulation_steps: 4
|
| 44 |
+
- total_train_batch_size: 64
|
| 45 |
+
- total_eval_batch_size: 32
|
| 46 |
+
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
|
| 47 |
+
- lr_scheduler_type: cosine
|
| 48 |
+
- lr_scheduler_warmup_ratio: 0.1
|
| 49 |
+
- num_epochs: 10
|
| 50 |
+
|
| 51 |
+
### Training results
|
| 52 |
+
|
| 53 |
+
|
| 54 |
+
|
| 55 |
+
### Framework versions
|
| 56 |
+
|
| 57 |
+
- Transformers 4.44.0
|
| 58 |
+
- Pytorch 2.1.2
|
| 59 |
+
- Datasets 3.0.0
|
| 60 |
+
- Tokenizers 0.19.1
|
all_results.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 10.0,
|
| 3 |
+
"total_flos": 0.0,
|
| 4 |
+
"train_loss": 0.048202857348136605,
|
| 5 |
+
"train_runtime": 22140.6758,
|
| 6 |
+
"train_samples": 1600,
|
| 7 |
+
"train_samples_per_second": 0.723,
|
| 8 |
+
"train_steps_per_second": 0.011
|
| 9 |
+
}
|
generation_config.json
ADDED
|
@@ -0,0 +1,12 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"bos_token_id": 128000,
|
| 3 |
+
"do_sample": true,
|
| 4 |
+
"eos_token_id": [
|
| 5 |
+
128001,
|
| 6 |
+
128009
|
| 7 |
+
],
|
| 8 |
+
"max_length": 4096,
|
| 9 |
+
"temperature": 0.6,
|
| 10 |
+
"top_p": 0.9,
|
| 11 |
+
"transformers_version": "4.44.0"
|
| 12 |
+
}
|
train_results.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 10.0,
|
| 3 |
+
"total_flos": 0.0,
|
| 4 |
+
"train_loss": 0.048202857348136605,
|
| 5 |
+
"train_runtime": 22140.6758,
|
| 6 |
+
"train_samples": 1600,
|
| 7 |
+
"train_samples_per_second": 0.723,
|
| 8 |
+
"train_steps_per_second": 0.011
|
| 9 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,432 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 10.0,
|
| 5 |
+
"eval_steps": 1000,
|
| 6 |
+
"global_step": 250,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.04,
|
| 13 |
+
"grad_norm": 5.49031856887175,
|
| 14 |
+
"learning_rate": 4.0000000000000003e-07,
|
| 15 |
+
"logits/chosen": -1.73323655128479,
|
| 16 |
+
"logits/rejected": -1.963712453842163,
|
| 17 |
+
"logps/chosen": -64.71795654296875,
|
| 18 |
+
"logps/rejected": -92.56527709960938,
|
| 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": 0.4,
|
| 28 |
+
"grad_norm": 1.9493782971816778,
|
| 29 |
+
"learning_rate": 4.000000000000001e-06,
|
| 30 |
+
"logits/chosen": -1.7276591062545776,
|
| 31 |
+
"logits/rejected": -1.90531325340271,
|
| 32 |
+
"logps/chosen": -72.89968872070312,
|
| 33 |
+
"logps/rejected": -117.47108459472656,
|
| 34 |
+
"loss": 0.6205,
|
| 35 |
+
"rewards/accuracies": 0.6180555820465088,
|
| 36 |
+
"rewards/chosen": -0.09507845342159271,
|
| 37 |
+
"rewards/margins": 0.2017170786857605,
|
| 38 |
+
"rewards/rejected": -0.296795517206192,
|
| 39 |
+
"step": 10
|
| 40 |
+
},
|
| 41 |
+
{
|
| 42 |
+
"epoch": 0.8,
|
| 43 |
+
"grad_norm": 3.7554212607527964,
|
| 44 |
+
"learning_rate": 8.000000000000001e-06,
|
| 45 |
+
"logits/chosen": -1.4812278747558594,
|
| 46 |
+
"logits/rejected": -1.7664066553115845,
|
| 47 |
+
"logps/chosen": -64.52765655517578,
|
| 48 |
+
"logps/rejected": -272.7065734863281,
|
| 49 |
+
"loss": 0.3953,
|
| 50 |
+
"rewards/accuracies": 0.8374999761581421,
|
| 51 |
+
"rewards/chosen": -0.0347316637635231,
|
| 52 |
+
"rewards/margins": 1.7828114032745361,
|
| 53 |
+
"rewards/rejected": -1.8175432682037354,
|
| 54 |
+
"step": 20
|
| 55 |
+
},
|
| 56 |
+
{
|
| 57 |
+
"epoch": 1.2,
|
| 58 |
+
"grad_norm": 6.377266655785215,
|
| 59 |
+
"learning_rate": 9.987820251299121e-06,
|
| 60 |
+
"logits/chosen": -1.6151340007781982,
|
| 61 |
+
"logits/rejected": -1.9460217952728271,
|
| 62 |
+
"logps/chosen": -143.10971069335938,
|
| 63 |
+
"logps/rejected": -631.4593505859375,
|
| 64 |
+
"loss": 0.1279,
|
| 65 |
+
"rewards/accuracies": 0.981249988079071,
|
| 66 |
+
"rewards/chosen": -0.8146063685417175,
|
| 67 |
+
"rewards/margins": 4.596449851989746,
|
| 68 |
+
"rewards/rejected": -5.4110565185546875,
|
| 69 |
+
"step": 30
|
| 70 |
+
},
|
| 71 |
+
{
|
| 72 |
+
"epoch": 1.6,
|
| 73 |
+
"grad_norm": 0.5750895192621723,
|
| 74 |
+
"learning_rate": 9.890738003669029e-06,
|
| 75 |
+
"logits/chosen": -1.6781848669052124,
|
| 76 |
+
"logits/rejected": -2.1725101470947266,
|
| 77 |
+
"logps/chosen": -153.36740112304688,
|
| 78 |
+
"logps/rejected": -981.5718994140625,
|
| 79 |
+
"loss": 0.0112,
|
| 80 |
+
"rewards/accuracies": 1.0,
|
| 81 |
+
"rewards/chosen": -0.9592069387435913,
|
| 82 |
+
"rewards/margins": 7.915135383605957,
|
| 83 |
+
"rewards/rejected": -8.87434196472168,
|
| 84 |
+
"step": 40
|
| 85 |
+
},
|
| 86 |
+
{
|
| 87 |
+
"epoch": 2.0,
|
| 88 |
+
"grad_norm": 15.42824961576695,
|
| 89 |
+
"learning_rate": 9.698463103929542e-06,
|
| 90 |
+
"logits/chosen": -2.032032012939453,
|
| 91 |
+
"logits/rejected": -2.422545909881592,
|
| 92 |
+
"logps/chosen": -291.60833740234375,
|
| 93 |
+
"logps/rejected": -1471.12451171875,
|
| 94 |
+
"loss": 0.0265,
|
| 95 |
+
"rewards/accuracies": 0.9937499761581421,
|
| 96 |
+
"rewards/chosen": -2.294036388397217,
|
| 97 |
+
"rewards/margins": 11.53666877746582,
|
| 98 |
+
"rewards/rejected": -13.830705642700195,
|
| 99 |
+
"step": 50
|
| 100 |
+
},
|
| 101 |
+
{
|
| 102 |
+
"epoch": 2.4,
|
| 103 |
+
"grad_norm": 0.04134383974122578,
|
| 104 |
+
"learning_rate": 9.414737964294636e-06,
|
| 105 |
+
"logits/chosen": -1.654057264328003,
|
| 106 |
+
"logits/rejected": -2.1446919441223145,
|
| 107 |
+
"logps/chosen": -305.9764709472656,
|
| 108 |
+
"logps/rejected": -1340.496826171875,
|
| 109 |
+
"loss": 0.0101,
|
| 110 |
+
"rewards/accuracies": 1.0,
|
| 111 |
+
"rewards/chosen": -2.476349353790283,
|
| 112 |
+
"rewards/margins": 9.981760025024414,
|
| 113 |
+
"rewards/rejected": -12.458109855651855,
|
| 114 |
+
"step": 60
|
| 115 |
+
},
|
| 116 |
+
{
|
| 117 |
+
"epoch": 2.8,
|
| 118 |
+
"grad_norm": 0.08981946731483229,
|
| 119 |
+
"learning_rate": 9.045084971874738e-06,
|
| 120 |
+
"logits/chosen": -1.8004175424575806,
|
| 121 |
+
"logits/rejected": -2.220939874649048,
|
| 122 |
+
"logps/chosen": -351.9838562011719,
|
| 123 |
+
"logps/rejected": -1534.653564453125,
|
| 124 |
+
"loss": 0.0004,
|
| 125 |
+
"rewards/accuracies": 1.0,
|
| 126 |
+
"rewards/chosen": -2.8868424892425537,
|
| 127 |
+
"rewards/margins": 11.563664436340332,
|
| 128 |
+
"rewards/rejected": -14.450506210327148,
|
| 129 |
+
"step": 70
|
| 130 |
+
},
|
| 131 |
+
{
|
| 132 |
+
"epoch": 3.2,
|
| 133 |
+
"grad_norm": 0.004562484598212617,
|
| 134 |
+
"learning_rate": 8.596699001693257e-06,
|
| 135 |
+
"logits/chosen": -1.955934762954712,
|
| 136 |
+
"logits/rejected": -2.4210100173950195,
|
| 137 |
+
"logps/chosen": -273.1143493652344,
|
| 138 |
+
"logps/rejected": -1595.3917236328125,
|
| 139 |
+
"loss": 0.0003,
|
| 140 |
+
"rewards/accuracies": 1.0,
|
| 141 |
+
"rewards/chosen": -2.1271727085113525,
|
| 142 |
+
"rewards/margins": 12.920974731445312,
|
| 143 |
+
"rewards/rejected": -15.048149108886719,
|
| 144 |
+
"step": 80
|
| 145 |
+
},
|
| 146 |
+
{
|
| 147 |
+
"epoch": 3.6,
|
| 148 |
+
"grad_norm": 0.1380036178242026,
|
| 149 |
+
"learning_rate": 8.078307376628292e-06,
|
| 150 |
+
"logits/chosen": -2.363739252090454,
|
| 151 |
+
"logits/rejected": -2.916693925857544,
|
| 152 |
+
"logps/chosen": -290.6210021972656,
|
| 153 |
+
"logps/rejected": -1820.208251953125,
|
| 154 |
+
"loss": 0.0001,
|
| 155 |
+
"rewards/accuracies": 1.0,
|
| 156 |
+
"rewards/chosen": -2.2858262062072754,
|
| 157 |
+
"rewards/margins": 15.026013374328613,
|
| 158 |
+
"rewards/rejected": -17.311840057373047,
|
| 159 |
+
"step": 90
|
| 160 |
+
},
|
| 161 |
+
{
|
| 162 |
+
"epoch": 4.0,
|
| 163 |
+
"grad_norm": 0.027548373376636768,
|
| 164 |
+
"learning_rate": 7.500000000000001e-06,
|
| 165 |
+
"logits/chosen": -1.8597825765609741,
|
| 166 |
+
"logits/rejected": -2.5159268379211426,
|
| 167 |
+
"logps/chosen": -259.3269348144531,
|
| 168 |
+
"logps/rejected": -1694.156005859375,
|
| 169 |
+
"loss": 0.0036,
|
| 170 |
+
"rewards/accuracies": 1.0,
|
| 171 |
+
"rewards/chosen": -1.9733333587646484,
|
| 172 |
+
"rewards/margins": 14.052156448364258,
|
| 173 |
+
"rewards/rejected": -16.025489807128906,
|
| 174 |
+
"step": 100
|
| 175 |
+
},
|
| 176 |
+
{
|
| 177 |
+
"epoch": 4.4,
|
| 178 |
+
"grad_norm": 0.002852087112507335,
|
| 179 |
+
"learning_rate": 6.873032967079562e-06,
|
| 180 |
+
"logits/chosen": -1.1590913534164429,
|
| 181 |
+
"logits/rejected": -1.9738283157348633,
|
| 182 |
+
"logps/chosen": -243.7049560546875,
|
| 183 |
+
"logps/rejected": -1585.557373046875,
|
| 184 |
+
"loss": 0.0008,
|
| 185 |
+
"rewards/accuracies": 1.0,
|
| 186 |
+
"rewards/chosen": -1.8130344152450562,
|
| 187 |
+
"rewards/margins": 13.158452033996582,
|
| 188 |
+
"rewards/rejected": -14.97148609161377,
|
| 189 |
+
"step": 110
|
| 190 |
+
},
|
| 191 |
+
{
|
| 192 |
+
"epoch": 4.8,
|
| 193 |
+
"grad_norm": 0.050302608930815555,
|
| 194 |
+
"learning_rate": 6.209609477998339e-06,
|
| 195 |
+
"logits/chosen": -1.1821445226669312,
|
| 196 |
+
"logits/rejected": -2.0527586936950684,
|
| 197 |
+
"logps/chosen": -265.8364562988281,
|
| 198 |
+
"logps/rejected": -1697.0533447265625,
|
| 199 |
+
"loss": 0.0002,
|
| 200 |
+
"rewards/accuracies": 1.0,
|
| 201 |
+
"rewards/chosen": -2.049001693725586,
|
| 202 |
+
"rewards/margins": 14.01880168914795,
|
| 203 |
+
"rewards/rejected": -16.06780242919922,
|
| 204 |
+
"step": 120
|
| 205 |
+
},
|
| 206 |
+
{
|
| 207 |
+
"epoch": 5.2,
|
| 208 |
+
"grad_norm": 0.002082230483938349,
|
| 209 |
+
"learning_rate": 5.522642316338268e-06,
|
| 210 |
+
"logits/chosen": -1.4268571138381958,
|
| 211 |
+
"logits/rejected": -2.196958303451538,
|
| 212 |
+
"logps/chosen": -261.50567626953125,
|
| 213 |
+
"logps/rejected": -1654.785888671875,
|
| 214 |
+
"loss": 0.0002,
|
| 215 |
+
"rewards/accuracies": 1.0,
|
| 216 |
+
"rewards/chosen": -2.0303101539611816,
|
| 217 |
+
"rewards/margins": 13.576850891113281,
|
| 218 |
+
"rewards/rejected": -15.607162475585938,
|
| 219 |
+
"step": 130
|
| 220 |
+
},
|
| 221 |
+
{
|
| 222 |
+
"epoch": 5.6,
|
| 223 |
+
"grad_norm": 0.0019893945992160438,
|
| 224 |
+
"learning_rate": 4.825502516487497e-06,
|
| 225 |
+
"logits/chosen": -1.7991822957992554,
|
| 226 |
+
"logits/rejected": -2.550363063812256,
|
| 227 |
+
"logps/chosen": -319.41717529296875,
|
| 228 |
+
"logps/rejected": -1847.844970703125,
|
| 229 |
+
"loss": 0.0001,
|
| 230 |
+
"rewards/accuracies": 1.0,
|
| 231 |
+
"rewards/chosen": -2.583193302154541,
|
| 232 |
+
"rewards/margins": 14.985641479492188,
|
| 233 |
+
"rewards/rejected": -17.568836212158203,
|
| 234 |
+
"step": 140
|
| 235 |
+
},
|
| 236 |
+
{
|
| 237 |
+
"epoch": 6.0,
|
| 238 |
+
"grad_norm": 0.0015741333192918123,
|
| 239 |
+
"learning_rate": 4.131759111665349e-06,
|
| 240 |
+
"logits/chosen": -1.9366722106933594,
|
| 241 |
+
"logits/rejected": -2.7011678218841553,
|
| 242 |
+
"logps/chosen": -317.91302490234375,
|
| 243 |
+
"logps/rejected": -1850.040771484375,
|
| 244 |
+
"loss": 0.0001,
|
| 245 |
+
"rewards/accuracies": 1.0,
|
| 246 |
+
"rewards/chosen": -2.5857417583465576,
|
| 247 |
+
"rewards/margins": 15.00117015838623,
|
| 248 |
+
"rewards/rejected": -17.586910247802734,
|
| 249 |
+
"step": 150
|
| 250 |
+
},
|
| 251 |
+
{
|
| 252 |
+
"epoch": 6.4,
|
| 253 |
+
"grad_norm": 0.0011046666078130404,
|
| 254 |
+
"learning_rate": 3.4549150281252635e-06,
|
| 255 |
+
"logits/chosen": -1.9223568439483643,
|
| 256 |
+
"logits/rejected": -2.68572735786438,
|
| 257 |
+
"logps/chosen": -357.874755859375,
|
| 258 |
+
"logps/rejected": -2010.9976806640625,
|
| 259 |
+
"loss": 0.0001,
|
| 260 |
+
"rewards/accuracies": 1.0,
|
| 261 |
+
"rewards/chosen": -2.9552831649780273,
|
| 262 |
+
"rewards/margins": 16.265609741210938,
|
| 263 |
+
"rewards/rejected": -19.22089195251465,
|
| 264 |
+
"step": 160
|
| 265 |
+
},
|
| 266 |
+
{
|
| 267 |
+
"epoch": 6.8,
|
| 268 |
+
"grad_norm": 0.0008850683483477801,
|
| 269 |
+
"learning_rate": 2.8081442660546126e-06,
|
| 270 |
+
"logits/chosen": -1.939117431640625,
|
| 271 |
+
"logits/rejected": -2.6780524253845215,
|
| 272 |
+
"logps/chosen": -341.6114196777344,
|
| 273 |
+
"logps/rejected": -1941.843017578125,
|
| 274 |
+
"loss": 0.0,
|
| 275 |
+
"rewards/accuracies": 1.0,
|
| 276 |
+
"rewards/chosen": -2.7991585731506348,
|
| 277 |
+
"rewards/margins": 15.720464706420898,
|
| 278 |
+
"rewards/rejected": -18.519622802734375,
|
| 279 |
+
"step": 170
|
| 280 |
+
},
|
| 281 |
+
{
|
| 282 |
+
"epoch": 7.2,
|
| 283 |
+
"grad_norm": 0.0006990799912761066,
|
| 284 |
+
"learning_rate": 2.204035482646267e-06,
|
| 285 |
+
"logits/chosen": -2.032839298248291,
|
| 286 |
+
"logits/rejected": -2.8106019496917725,
|
| 287 |
+
"logps/chosen": -401.89849853515625,
|
| 288 |
+
"logps/rejected": -2151.74072265625,
|
| 289 |
+
"loss": 0.0,
|
| 290 |
+
"rewards/accuracies": 1.0,
|
| 291 |
+
"rewards/chosen": -3.3739535808563232,
|
| 292 |
+
"rewards/margins": 17.275264739990234,
|
| 293 |
+
"rewards/rejected": -20.649219512939453,
|
| 294 |
+
"step": 180
|
| 295 |
+
},
|
| 296 |
+
{
|
| 297 |
+
"epoch": 7.6,
|
| 298 |
+
"grad_norm": 0.0009810237591607713,
|
| 299 |
+
"learning_rate": 1.6543469682057105e-06,
|
| 300 |
+
"logits/chosen": -2.005420446395874,
|
| 301 |
+
"logits/rejected": -2.788722038269043,
|
| 302 |
+
"logps/chosen": -371.88824462890625,
|
| 303 |
+
"logps/rejected": -2029.203125,
|
| 304 |
+
"loss": 0.0,
|
| 305 |
+
"rewards/accuracies": 1.0,
|
| 306 |
+
"rewards/chosen": -3.0968661308288574,
|
| 307 |
+
"rewards/margins": 16.302059173583984,
|
| 308 |
+
"rewards/rejected": -19.398929595947266,
|
| 309 |
+
"step": 190
|
| 310 |
+
},
|
| 311 |
+
{
|
| 312 |
+
"epoch": 8.0,
|
| 313 |
+
"grad_norm": 0.005038481428283606,
|
| 314 |
+
"learning_rate": 1.1697777844051105e-06,
|
| 315 |
+
"logits/chosen": -1.6506569385528564,
|
| 316 |
+
"logits/rejected": -2.4032554626464844,
|
| 317 |
+
"logps/chosen": -272.77264404296875,
|
| 318 |
+
"logps/rejected": -1795.190185546875,
|
| 319 |
+
"loss": 0.0,
|
| 320 |
+
"rewards/accuracies": 1.0,
|
| 321 |
+
"rewards/chosen": -2.1385960578918457,
|
| 322 |
+
"rewards/margins": 14.876760482788086,
|
| 323 |
+
"rewards/rejected": -17.01535987854004,
|
| 324 |
+
"step": 200
|
| 325 |
+
},
|
| 326 |
+
{
|
| 327 |
+
"epoch": 8.4,
|
| 328 |
+
"grad_norm": 0.0007269378594182772,
|
| 329 |
+
"learning_rate": 7.597595192178702e-07,
|
| 330 |
+
"logits/chosen": -1.87616765499115,
|
| 331 |
+
"logits/rejected": -2.6102375984191895,
|
| 332 |
+
"logps/chosen": -321.3392639160156,
|
| 333 |
+
"logps/rejected": -1848.580810546875,
|
| 334 |
+
"loss": 0.0,
|
| 335 |
+
"rewards/accuracies": 1.0,
|
| 336 |
+
"rewards/chosen": -2.6000113487243652,
|
| 337 |
+
"rewards/margins": 14.976901054382324,
|
| 338 |
+
"rewards/rejected": -17.576915740966797,
|
| 339 |
+
"step": 210
|
| 340 |
+
},
|
| 341 |
+
{
|
| 342 |
+
"epoch": 8.8,
|
| 343 |
+
"grad_norm": 0.0007139184876414333,
|
| 344 |
+
"learning_rate": 4.322727117869951e-07,
|
| 345 |
+
"logits/chosen": -1.9712779521942139,
|
| 346 |
+
"logits/rejected": -2.749927043914795,
|
| 347 |
+
"logps/chosen": -361.2076416015625,
|
| 348 |
+
"logps/rejected": -1984.336669921875,
|
| 349 |
+
"loss": 0.0,
|
| 350 |
+
"rewards/accuracies": 1.0,
|
| 351 |
+
"rewards/chosen": -2.9964027404785156,
|
| 352 |
+
"rewards/margins": 15.949694633483887,
|
| 353 |
+
"rewards/rejected": -18.946096420288086,
|
| 354 |
+
"step": 220
|
| 355 |
+
},
|
| 356 |
+
{
|
| 357 |
+
"epoch": 9.2,
|
| 358 |
+
"grad_norm": 0.0009109490304217519,
|
| 359 |
+
"learning_rate": 1.9369152030840553e-07,
|
| 360 |
+
"logits/chosen": -1.8861472606658936,
|
| 361 |
+
"logits/rejected": -2.6229307651519775,
|
| 362 |
+
"logps/chosen": -340.4961853027344,
|
| 363 |
+
"logps/rejected": -1894.4420166015625,
|
| 364 |
+
"loss": 0.0,
|
| 365 |
+
"rewards/accuracies": 1.0,
|
| 366 |
+
"rewards/chosen": -2.806462287902832,
|
| 367 |
+
"rewards/margins": 15.230853080749512,
|
| 368 |
+
"rewards/rejected": -18.037317276000977,
|
| 369 |
+
"step": 230
|
| 370 |
+
},
|
| 371 |
+
{
|
| 372 |
+
"epoch": 9.6,
|
| 373 |
+
"grad_norm": 0.0007160536583349626,
|
| 374 |
+
"learning_rate": 4.865965629214819e-08,
|
| 375 |
+
"logits/chosen": -1.8993374109268188,
|
| 376 |
+
"logits/rejected": -2.6610684394836426,
|
| 377 |
+
"logps/chosen": -320.7115783691406,
|
| 378 |
+
"logps/rejected": -1872.178466796875,
|
| 379 |
+
"loss": 0.0,
|
| 380 |
+
"rewards/accuracies": 1.0,
|
| 381 |
+
"rewards/chosen": -2.6092982292175293,
|
| 382 |
+
"rewards/margins": 15.210273742675781,
|
| 383 |
+
"rewards/rejected": -17.81957244873047,
|
| 384 |
+
"step": 240
|
| 385 |
+
},
|
| 386 |
+
{
|
| 387 |
+
"epoch": 10.0,
|
| 388 |
+
"grad_norm": 0.0006100741232287624,
|
| 389 |
+
"learning_rate": 0.0,
|
| 390 |
+
"logits/chosen": -1.9146867990493774,
|
| 391 |
+
"logits/rejected": -2.670241117477417,
|
| 392 |
+
"logps/chosen": -317.877685546875,
|
| 393 |
+
"logps/rejected": -1852.078857421875,
|
| 394 |
+
"loss": 0.0,
|
| 395 |
+
"rewards/accuracies": 1.0,
|
| 396 |
+
"rewards/chosen": -2.5683541297912598,
|
| 397 |
+
"rewards/margins": 15.033134460449219,
|
| 398 |
+
"rewards/rejected": -17.60148811340332,
|
| 399 |
+
"step": 250
|
| 400 |
+
},
|
| 401 |
+
{
|
| 402 |
+
"epoch": 10.0,
|
| 403 |
+
"step": 250,
|
| 404 |
+
"total_flos": 0.0,
|
| 405 |
+
"train_loss": 0.048202857348136605,
|
| 406 |
+
"train_runtime": 22140.6758,
|
| 407 |
+
"train_samples_per_second": 0.723,
|
| 408 |
+
"train_steps_per_second": 0.011
|
| 409 |
+
}
|
| 410 |
+
],
|
| 411 |
+
"logging_steps": 10,
|
| 412 |
+
"max_steps": 250,
|
| 413 |
+
"num_input_tokens_seen": 0,
|
| 414 |
+
"num_train_epochs": 10,
|
| 415 |
+
"save_steps": 1000,
|
| 416 |
+
"stateful_callbacks": {
|
| 417 |
+
"TrainerControl": {
|
| 418 |
+
"args": {
|
| 419 |
+
"should_epoch_stop": false,
|
| 420 |
+
"should_evaluate": false,
|
| 421 |
+
"should_log": false,
|
| 422 |
+
"should_save": true,
|
| 423 |
+
"should_training_stop": true
|
| 424 |
+
},
|
| 425 |
+
"attributes": {}
|
| 426 |
+
}
|
| 427 |
+
},
|
| 428 |
+
"total_flos": 0.0,
|
| 429 |
+
"train_batch_size": 4,
|
| 430 |
+
"trial_name": null,
|
| 431 |
+
"trial_params": null
|
| 432 |
+
}
|