Model save
Browse files- README.md +83 -0
- adapter_model.safetensors +1 -1
- all_results.json +21 -0
- eval_results.json +16 -0
- runs/Mar04_23-03-52_gpu4-119-4/events.out.tfevents.1709553990.gpu4-119-4.2645530.0 +2 -2
- runs/Mar04_23-03-52_gpu4-119-4/events.out.tfevents.1709572444.gpu4-119-4.2645530.1 +3 -0
- train_results.json +8 -0
- trainer_state.json +1518 -0
README.md
ADDED
|
@@ -0,0 +1,83 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
license: mit
|
| 3 |
+
library_name: peft
|
| 4 |
+
tags:
|
| 5 |
+
- trl
|
| 6 |
+
- dpo
|
| 7 |
+
- generated_from_trainer
|
| 8 |
+
base_model: microsoft/phi-2
|
| 9 |
+
model-index:
|
| 10 |
+
- name: phi-2-gpo-ultrafeedback-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 |
+
# phi-2-gpo-ultrafeedback-lora
|
| 18 |
+
|
| 19 |
+
This model is a fine-tuned version of [microsoft/phi-2](https://huggingface.co/microsoft/phi-2) on the None dataset.
|
| 20 |
+
It achieves the following results on the evaluation set:
|
| 21 |
+
- Loss: 0.0021
|
| 22 |
+
- Rewards/chosen: -0.0083
|
| 23 |
+
- Rewards/rejected: -0.0184
|
| 24 |
+
- Rewards/accuracies: 0.6920
|
| 25 |
+
- Rewards/margins: 0.0101
|
| 26 |
+
- Logps/rejected: -233.2711
|
| 27 |
+
- Logps/chosen: -261.0694
|
| 28 |
+
- Logits/rejected: 0.8833
|
| 29 |
+
- Logits/chosen: 0.7809
|
| 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-06
|
| 49 |
+
- train_batch_size: 4
|
| 50 |
+
- eval_batch_size: 4
|
| 51 |
+
- seed: 42
|
| 52 |
+
- distributed_type: multi-GPU
|
| 53 |
+
- num_devices: 4
|
| 54 |
+
- gradient_accumulation_steps: 4
|
| 55 |
+
- total_train_batch_size: 64
|
| 56 |
+
- total_eval_batch_size: 16
|
| 57 |
+
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
|
| 58 |
+
- lr_scheduler_type: cosine
|
| 59 |
+
- lr_scheduler_warmup_ratio: 0.1
|
| 60 |
+
- num_epochs: 2
|
| 61 |
+
|
| 62 |
+
### Training results
|
| 63 |
+
|
| 64 |
+
| Training Loss | Epoch | Step | Validation Loss | Rewards/chosen | Rewards/rejected | Rewards/accuracies | Rewards/margins | Logps/rejected | Logps/chosen | Logits/rejected | Logits/chosen |
|
| 65 |
+
|:-------------:|:-----:|:----:|:---------------:|:--------------:|:----------------:|:------------------:|:---------------:|:--------------:|:------------:|:---------------:|:-------------:|
|
| 66 |
+
| 0.0026 | 0.21 | 100 | 0.0025 | 0.0001 | -0.0005 | 0.5080 | 0.0006 | -231.4896 | -260.2373 | 0.9175 | 0.8151 |
|
| 67 |
+
| 0.0023 | 0.42 | 200 | 0.0023 | -0.0015 | -0.0068 | 0.6560 | 0.0053 | -232.1152 | -260.3932 | 0.9120 | 0.8092 |
|
| 68 |
+
| 0.0022 | 0.63 | 300 | 0.0022 | -0.0067 | -0.0141 | 0.6700 | 0.0073 | -232.8447 | -260.9179 | 0.9022 | 0.7992 |
|
| 69 |
+
| 0.0021 | 0.84 | 400 | 0.0022 | -0.0092 | -0.0178 | 0.6640 | 0.0086 | -233.2157 | -261.1620 | 0.8914 | 0.7884 |
|
| 70 |
+
| 0.0022 | 1.05 | 500 | 0.0021 | -0.0094 | -0.0193 | 0.7100 | 0.0098 | -233.3614 | -261.1852 | 0.8853 | 0.7821 |
|
| 71 |
+
| 0.002 | 1.26 | 600 | 0.0021 | -0.0088 | -0.0185 | 0.6940 | 0.0097 | -233.2843 | -261.1207 | 0.8840 | 0.7815 |
|
| 72 |
+
| 0.0021 | 1.47 | 700 | 0.0021 | -0.0083 | -0.0182 | 0.7000 | 0.0099 | -233.2560 | -261.0788 | 0.8816 | 0.7790 |
|
| 73 |
+
| 0.0021 | 1.67 | 800 | 0.0021 | -0.0082 | -0.0184 | 0.6940 | 0.0102 | -233.2740 | -261.0643 | 0.8811 | 0.7781 |
|
| 74 |
+
| 0.0021 | 1.88 | 900 | 0.0021 | -0.0085 | -0.0178 | 0.6900 | 0.0093 | -233.2118 | -261.0922 | 0.8833 | 0.7806 |
|
| 75 |
+
|
| 76 |
+
|
| 77 |
+
### Framework versions
|
| 78 |
+
|
| 79 |
+
- PEFT 0.7.1
|
| 80 |
+
- Transformers 4.36.2
|
| 81 |
+
- Pytorch 2.1.2+cu118
|
| 82 |
+
- Datasets 2.14.6
|
| 83 |
+
- Tokenizers 0.15.2
|
adapter_model.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 41977616
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:d0dea53272310862afda3712da656b50c1d9a7ad7a46f0642635168a85f6d5a0
|
| 3 |
size 41977616
|
all_results.json
ADDED
|
@@ -0,0 +1,21 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 2.0,
|
| 3 |
+
"eval_logits/chosen": 0.7808946371078491,
|
| 4 |
+
"eval_logits/rejected": 0.8833128213882446,
|
| 5 |
+
"eval_logps/chosen": -261.0694274902344,
|
| 6 |
+
"eval_logps/rejected": -233.27114868164062,
|
| 7 |
+
"eval_loss": 0.0021080097649246454,
|
| 8 |
+
"eval_rewards/accuracies": 0.6919999718666077,
|
| 9 |
+
"eval_rewards/chosen": -0.008252721279859543,
|
| 10 |
+
"eval_rewards/margins": 0.01009758934378624,
|
| 11 |
+
"eval_rewards/rejected": -0.018350308761000633,
|
| 12 |
+
"eval_runtime": 325.1898,
|
| 13 |
+
"eval_samples": 2000,
|
| 14 |
+
"eval_samples_per_second": 6.15,
|
| 15 |
+
"eval_steps_per_second": 0.384,
|
| 16 |
+
"train_loss": 0.0021909422920118682,
|
| 17 |
+
"train_runtime": 18127.9992,
|
| 18 |
+
"train_samples": 30567,
|
| 19 |
+
"train_samples_per_second": 3.372,
|
| 20 |
+
"train_steps_per_second": 0.053
|
| 21 |
+
}
|
eval_results.json
ADDED
|
@@ -0,0 +1,16 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 2.0,
|
| 3 |
+
"eval_logits/chosen": 0.7808946371078491,
|
| 4 |
+
"eval_logits/rejected": 0.8833128213882446,
|
| 5 |
+
"eval_logps/chosen": -261.0694274902344,
|
| 6 |
+
"eval_logps/rejected": -233.27114868164062,
|
| 7 |
+
"eval_loss": 0.0021080097649246454,
|
| 8 |
+
"eval_rewards/accuracies": 0.6919999718666077,
|
| 9 |
+
"eval_rewards/chosen": -0.008252721279859543,
|
| 10 |
+
"eval_rewards/margins": 0.01009758934378624,
|
| 11 |
+
"eval_rewards/rejected": -0.018350308761000633,
|
| 12 |
+
"eval_runtime": 325.1898,
|
| 13 |
+
"eval_samples": 2000,
|
| 14 |
+
"eval_samples_per_second": 6.15,
|
| 15 |
+
"eval_steps_per_second": 0.384
|
| 16 |
+
}
|
runs/Mar04_23-03-52_gpu4-119-4/events.out.tfevents.1709553990.gpu4-119-4.2645530.0
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
-
size
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:86e7d65dd1ecab0062868e9eb0fa64b5d449a74a3b81d0fccc5cbbfc683b08ee
|
| 3 |
+
size 72564
|
runs/Mar04_23-03-52_gpu4-119-4/events.out.tfevents.1709572444.gpu4-119-4.2645530.1
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:8536b7640e441d92bc78ca3e370ef6cd1c50bece09183ec2c87d1de81f9f887e
|
| 3 |
+
size 828
|
train_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 2.0,
|
| 3 |
+
"train_loss": 0.0021909422920118682,
|
| 4 |
+
"train_runtime": 18127.9992,
|
| 5 |
+
"train_samples": 30567,
|
| 6 |
+
"train_samples_per_second": 3.372,
|
| 7 |
+
"train_steps_per_second": 0.053
|
| 8 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,1518 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 1.9968602825745683,
|
| 5 |
+
"eval_steps": 100,
|
| 6 |
+
"global_step": 954,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.0,
|
| 13 |
+
"learning_rate": 5.208333333333333e-08,
|
| 14 |
+
"logits/chosen": 0.952304482460022,
|
| 15 |
+
"logits/rejected": 0.5888463854789734,
|
| 16 |
+
"logps/chosen": -223.79486083984375,
|
| 17 |
+
"logps/rejected": -209.482666015625,
|
| 18 |
+
"loss": 0.0025,
|
| 19 |
+
"rewards/accuracies": 0.0,
|
| 20 |
+
"rewards/chosen": 0.0,
|
| 21 |
+
"rewards/margins": 0.0,
|
| 22 |
+
"rewards/rejected": 0.0,
|
| 23 |
+
"step": 1
|
| 24 |
+
},
|
| 25 |
+
{
|
| 26 |
+
"epoch": 0.02,
|
| 27 |
+
"learning_rate": 5.208333333333334e-07,
|
| 28 |
+
"logits/chosen": 0.8366600871086121,
|
| 29 |
+
"logits/rejected": 0.8544472455978394,
|
| 30 |
+
"logps/chosen": -236.2534942626953,
|
| 31 |
+
"logps/rejected": -221.8985137939453,
|
| 32 |
+
"loss": 0.0026,
|
| 33 |
+
"rewards/accuracies": 0.4583333432674408,
|
| 34 |
+
"rewards/chosen": 9.055635746335611e-05,
|
| 35 |
+
"rewards/margins": -0.00016146278358064592,
|
| 36 |
+
"rewards/rejected": 0.00025201926473528147,
|
| 37 |
+
"step": 10
|
| 38 |
+
},
|
| 39 |
+
{
|
| 40 |
+
"epoch": 0.04,
|
| 41 |
+
"learning_rate": 1.0416666666666667e-06,
|
| 42 |
+
"logits/chosen": 0.8331616520881653,
|
| 43 |
+
"logits/rejected": 0.9283801317214966,
|
| 44 |
+
"logps/chosen": -254.7387237548828,
|
| 45 |
+
"logps/rejected": -247.924560546875,
|
| 46 |
+
"loss": 0.0025,
|
| 47 |
+
"rewards/accuracies": 0.581250011920929,
|
| 48 |
+
"rewards/chosen": -0.0002719077165238559,
|
| 49 |
+
"rewards/margins": 0.000702966412063688,
|
| 50 |
+
"rewards/rejected": -0.0009748738375492394,
|
| 51 |
+
"step": 20
|
| 52 |
+
},
|
| 53 |
+
{
|
| 54 |
+
"epoch": 0.06,
|
| 55 |
+
"learning_rate": 1.5625e-06,
|
| 56 |
+
"logits/chosen": 0.8597829937934875,
|
| 57 |
+
"logits/rejected": 0.9174444079399109,
|
| 58 |
+
"logps/chosen": -260.46356201171875,
|
| 59 |
+
"logps/rejected": -232.0428924560547,
|
| 60 |
+
"loss": 0.0025,
|
| 61 |
+
"rewards/accuracies": 0.518750011920929,
|
| 62 |
+
"rewards/chosen": 0.0010247982572764158,
|
| 63 |
+
"rewards/margins": 0.0011705085635185242,
|
| 64 |
+
"rewards/rejected": -0.0001457103790016845,
|
| 65 |
+
"step": 30
|
| 66 |
+
},
|
| 67 |
+
{
|
| 68 |
+
"epoch": 0.08,
|
| 69 |
+
"learning_rate": 2.0833333333333334e-06,
|
| 70 |
+
"logits/chosen": 0.8120288848876953,
|
| 71 |
+
"logits/rejected": 0.9034429788589478,
|
| 72 |
+
"logps/chosen": -280.2599792480469,
|
| 73 |
+
"logps/rejected": -228.59304809570312,
|
| 74 |
+
"loss": 0.0025,
|
| 75 |
+
"rewards/accuracies": 0.4625000059604645,
|
| 76 |
+
"rewards/chosen": -0.0004804997588507831,
|
| 77 |
+
"rewards/margins": -2.5537923647789285e-05,
|
| 78 |
+
"rewards/rejected": -0.00045496178790926933,
|
| 79 |
+
"step": 40
|
| 80 |
+
},
|
| 81 |
+
{
|
| 82 |
+
"epoch": 0.1,
|
| 83 |
+
"learning_rate": 2.604166666666667e-06,
|
| 84 |
+
"logits/chosen": 0.856873631477356,
|
| 85 |
+
"logits/rejected": 0.936148464679718,
|
| 86 |
+
"logps/chosen": -257.70074462890625,
|
| 87 |
+
"logps/rejected": -219.18563842773438,
|
| 88 |
+
"loss": 0.0025,
|
| 89 |
+
"rewards/accuracies": 0.574999988079071,
|
| 90 |
+
"rewards/chosen": -2.7128717192681506e-05,
|
| 91 |
+
"rewards/margins": 0.0010264910524711013,
|
| 92 |
+
"rewards/rejected": -0.0010536197805777192,
|
| 93 |
+
"step": 50
|
| 94 |
+
},
|
| 95 |
+
{
|
| 96 |
+
"epoch": 0.13,
|
| 97 |
+
"learning_rate": 3.125e-06,
|
| 98 |
+
"logits/chosen": 0.8756526708602905,
|
| 99 |
+
"logits/rejected": 0.9026013612747192,
|
| 100 |
+
"logps/chosen": -237.04653930664062,
|
| 101 |
+
"logps/rejected": -237.41769409179688,
|
| 102 |
+
"loss": 0.0025,
|
| 103 |
+
"rewards/accuracies": 0.4749999940395355,
|
| 104 |
+
"rewards/chosen": -0.00045221406617201865,
|
| 105 |
+
"rewards/margins": -0.0005772784352302551,
|
| 106 |
+
"rewards/rejected": 0.0001250644854735583,
|
| 107 |
+
"step": 60
|
| 108 |
+
},
|
| 109 |
+
{
|
| 110 |
+
"epoch": 0.15,
|
| 111 |
+
"learning_rate": 3.6458333333333333e-06,
|
| 112 |
+
"logits/chosen": 0.8502365946769714,
|
| 113 |
+
"logits/rejected": 0.878851592540741,
|
| 114 |
+
"logps/chosen": -260.7863464355469,
|
| 115 |
+
"logps/rejected": -227.58700561523438,
|
| 116 |
+
"loss": 0.0025,
|
| 117 |
+
"rewards/accuracies": 0.4937500059604645,
|
| 118 |
+
"rewards/chosen": -0.000642338243778795,
|
| 119 |
+
"rewards/margins": -0.0005514883669093251,
|
| 120 |
+
"rewards/rejected": -9.084997873287648e-05,
|
| 121 |
+
"step": 70
|
| 122 |
+
},
|
| 123 |
+
{
|
| 124 |
+
"epoch": 0.17,
|
| 125 |
+
"learning_rate": 4.166666666666667e-06,
|
| 126 |
+
"logits/chosen": 0.8864189982414246,
|
| 127 |
+
"logits/rejected": 0.9194203615188599,
|
| 128 |
+
"logps/chosen": -251.07644653320312,
|
| 129 |
+
"logps/rejected": -231.24453735351562,
|
| 130 |
+
"loss": 0.0025,
|
| 131 |
+
"rewards/accuracies": 0.53125,
|
| 132 |
+
"rewards/chosen": 8.48618583404459e-05,
|
| 133 |
+
"rewards/margins": 0.00046581291826441884,
|
| 134 |
+
"rewards/rejected": -0.00038095106719993055,
|
| 135 |
+
"step": 80
|
| 136 |
+
},
|
| 137 |
+
{
|
| 138 |
+
"epoch": 0.19,
|
| 139 |
+
"learning_rate": 4.6875000000000004e-06,
|
| 140 |
+
"logits/chosen": 0.8617275953292847,
|
| 141 |
+
"logits/rejected": 0.9164407849311829,
|
| 142 |
+
"logps/chosen": -225.30056762695312,
|
| 143 |
+
"logps/rejected": -241.3493194580078,
|
| 144 |
+
"loss": 0.0025,
|
| 145 |
+
"rewards/accuracies": 0.543749988079071,
|
| 146 |
+
"rewards/chosen": -0.00032968426239676774,
|
| 147 |
+
"rewards/margins": 0.0004738254356198013,
|
| 148 |
+
"rewards/rejected": -0.0008035098435357213,
|
| 149 |
+
"step": 90
|
| 150 |
+
},
|
| 151 |
+
{
|
| 152 |
+
"epoch": 0.21,
|
| 153 |
+
"learning_rate": 4.999731868769027e-06,
|
| 154 |
+
"logits/chosen": 0.9239044189453125,
|
| 155 |
+
"logits/rejected": 0.9236629605293274,
|
| 156 |
+
"logps/chosen": -242.187744140625,
|
| 157 |
+
"logps/rejected": -221.4308624267578,
|
| 158 |
+
"loss": 0.0026,
|
| 159 |
+
"rewards/accuracies": 0.518750011920929,
|
| 160 |
+
"rewards/chosen": -0.0005193214165046811,
|
| 161 |
+
"rewards/margins": 0.00021600276522804052,
|
| 162 |
+
"rewards/rejected": -0.0007353241671808064,
|
| 163 |
+
"step": 100
|
| 164 |
+
},
|
| 165 |
+
{
|
| 166 |
+
"epoch": 0.21,
|
| 167 |
+
"eval_logits/chosen": 0.8150850534439087,
|
| 168 |
+
"eval_logits/rejected": 0.917453408241272,
|
| 169 |
+
"eval_logps/chosen": -260.2373046875,
|
| 170 |
+
"eval_logps/rejected": -231.48963928222656,
|
| 171 |
+
"eval_loss": 0.00248239329084754,
|
| 172 |
+
"eval_rewards/accuracies": 0.5080000162124634,
|
| 173 |
+
"eval_rewards/chosen": 6.857867265352979e-05,
|
| 174 |
+
"eval_rewards/margins": 0.0006035350379534066,
|
| 175 |
+
"eval_rewards/rejected": -0.0005349563434720039,
|
| 176 |
+
"eval_runtime": 327.6723,
|
| 177 |
+
"eval_samples_per_second": 6.104,
|
| 178 |
+
"eval_steps_per_second": 0.381,
|
| 179 |
+
"step": 100
|
| 180 |
+
},
|
| 181 |
+
{
|
| 182 |
+
"epoch": 0.23,
|
| 183 |
+
"learning_rate": 4.996716052911017e-06,
|
| 184 |
+
"logits/chosen": 0.8340710401535034,
|
| 185 |
+
"logits/rejected": 0.8745189905166626,
|
| 186 |
+
"logps/chosen": -264.0061950683594,
|
| 187 |
+
"logps/rejected": -219.55416870117188,
|
| 188 |
+
"loss": 0.0025,
|
| 189 |
+
"rewards/accuracies": 0.574999988079071,
|
| 190 |
+
"rewards/chosen": 0.00015970889944583178,
|
| 191 |
+
"rewards/margins": 0.0009235168108716607,
|
| 192 |
+
"rewards/rejected": -0.000763807853218168,
|
| 193 |
+
"step": 110
|
| 194 |
+
},
|
| 195 |
+
{
|
| 196 |
+
"epoch": 0.25,
|
| 197 |
+
"learning_rate": 4.9903533134293035e-06,
|
| 198 |
+
"logits/chosen": 0.8607433438301086,
|
| 199 |
+
"logits/rejected": 0.9696208238601685,
|
| 200 |
+
"logps/chosen": -254.9775390625,
|
| 201 |
+
"logps/rejected": -219.08029174804688,
|
| 202 |
+
"loss": 0.0025,
|
| 203 |
+
"rewards/accuracies": 0.512499988079071,
|
| 204 |
+
"rewards/chosen": -0.00045476845116354525,
|
| 205 |
+
"rewards/margins": 0.0009122647461481392,
|
| 206 |
+
"rewards/rejected": -0.001367033226415515,
|
| 207 |
+
"step": 120
|
| 208 |
+
},
|
| 209 |
+
{
|
| 210 |
+
"epoch": 0.27,
|
| 211 |
+
"learning_rate": 4.9806521797692184e-06,
|
| 212 |
+
"logits/chosen": 0.8772906064987183,
|
| 213 |
+
"logits/rejected": 0.8781763911247253,
|
| 214 |
+
"logps/chosen": -264.6737365722656,
|
| 215 |
+
"logps/rejected": -246.98770141601562,
|
| 216 |
+
"loss": 0.0025,
|
| 217 |
+
"rewards/accuracies": 0.5062500238418579,
|
| 218 |
+
"rewards/chosen": -0.00038236891850829124,
|
| 219 |
+
"rewards/margins": 0.0011148005723953247,
|
| 220 |
+
"rewards/rejected": -0.0014971692580729723,
|
| 221 |
+
"step": 130
|
| 222 |
+
},
|
| 223 |
+
{
|
| 224 |
+
"epoch": 0.29,
|
| 225 |
+
"learning_rate": 4.967625656594782e-06,
|
| 226 |
+
"logits/chosen": 0.8523051142692566,
|
| 227 |
+
"logits/rejected": 0.9305570721626282,
|
| 228 |
+
"logps/chosen": -222.01806640625,
|
| 229 |
+
"logps/rejected": -232.17919921875,
|
| 230 |
+
"loss": 0.0024,
|
| 231 |
+
"rewards/accuracies": 0.4937500059604645,
|
| 232 |
+
"rewards/chosen": -0.0008891393663361669,
|
| 233 |
+
"rewards/margins": 0.0006222378578968346,
|
| 234 |
+
"rewards/rejected": -0.0015113770496100187,
|
| 235 |
+
"step": 140
|
| 236 |
+
},
|
| 237 |
+
{
|
| 238 |
+
"epoch": 0.31,
|
| 239 |
+
"learning_rate": 4.95129120635556e-06,
|
| 240 |
+
"logits/chosen": 0.8730077743530273,
|
| 241 |
+
"logits/rejected": 0.9034315943717957,
|
| 242 |
+
"logps/chosen": -258.2915954589844,
|
| 243 |
+
"logps/rejected": -215.7738037109375,
|
| 244 |
+
"loss": 0.0024,
|
| 245 |
+
"rewards/accuracies": 0.5375000238418579,
|
| 246 |
+
"rewards/chosen": -0.0006225308170542121,
|
| 247 |
+
"rewards/margins": 0.002022756729274988,
|
| 248 |
+
"rewards/rejected": -0.0026452874299138784,
|
| 249 |
+
"step": 150
|
| 250 |
+
},
|
| 251 |
+
{
|
| 252 |
+
"epoch": 0.33,
|
| 253 |
+
"learning_rate": 4.93167072587771e-06,
|
| 254 |
+
"logits/chosen": 0.7767001390457153,
|
| 255 |
+
"logits/rejected": 0.845874011516571,
|
| 256 |
+
"logps/chosen": -257.7080383300781,
|
| 257 |
+
"logps/rejected": -250.625732421875,
|
| 258 |
+
"loss": 0.0024,
|
| 259 |
+
"rewards/accuracies": 0.574999988079071,
|
| 260 |
+
"rewards/chosen": -0.0005334648885764182,
|
| 261 |
+
"rewards/margins": 0.0022061350755393505,
|
| 262 |
+
"rewards/rejected": -0.0027396001387387514,
|
| 263 |
+
"step": 160
|
| 264 |
+
},
|
| 265 |
+
{
|
| 266 |
+
"epoch": 0.36,
|
| 267 |
+
"learning_rate": 4.908790517010637e-06,
|
| 268 |
+
"logits/chosen": 0.9272924661636353,
|
| 269 |
+
"logits/rejected": 0.9679857492446899,
|
| 270 |
+
"logps/chosen": -239.0647735595703,
|
| 271 |
+
"logps/rejected": -252.8076934814453,
|
| 272 |
+
"loss": 0.0024,
|
| 273 |
+
"rewards/accuracies": 0.5562499761581421,
|
| 274 |
+
"rewards/chosen": -0.002121446654200554,
|
| 275 |
+
"rewards/margins": 0.001315777888521552,
|
| 276 |
+
"rewards/rejected": -0.0034372243098914623,
|
| 277 |
+
"step": 170
|
| 278 |
+
},
|
| 279 |
+
{
|
| 280 |
+
"epoch": 0.38,
|
| 281 |
+
"learning_rate": 4.882681251368549e-06,
|
| 282 |
+
"logits/chosen": 0.8513296246528625,
|
| 283 |
+
"logits/rejected": 0.8445970416069031,
|
| 284 |
+
"logps/chosen": -270.7970886230469,
|
| 285 |
+
"logps/rejected": -256.8065185546875,
|
| 286 |
+
"loss": 0.0023,
|
| 287 |
+
"rewards/accuracies": 0.606249988079071,
|
| 288 |
+
"rewards/chosen": -0.00029190973145887256,
|
| 289 |
+
"rewards/margins": 0.0030645509250462055,
|
| 290 |
+
"rewards/rejected": -0.0033564604818820953,
|
| 291 |
+
"step": 180
|
| 292 |
+
},
|
| 293 |
+
{
|
| 294 |
+
"epoch": 0.4,
|
| 295 |
+
"learning_rate": 4.853377929214243e-06,
|
| 296 |
+
"logits/chosen": 0.7836719751358032,
|
| 297 |
+
"logits/rejected": 0.8565704226493835,
|
| 298 |
+
"logps/chosen": -252.17233276367188,
|
| 299 |
+
"logps/rejected": -244.7781219482422,
|
| 300 |
+
"loss": 0.0023,
|
| 301 |
+
"rewards/accuracies": 0.606249988079071,
|
| 302 |
+
"rewards/chosen": -0.001070442027412355,
|
| 303 |
+
"rewards/margins": 0.0032107688020914793,
|
| 304 |
+
"rewards/rejected": -0.004281210713088512,
|
| 305 |
+
"step": 190
|
| 306 |
+
},
|
| 307 |
+
{
|
| 308 |
+
"epoch": 0.42,
|
| 309 |
+
"learning_rate": 4.8209198325401815e-06,
|
| 310 |
+
"logits/chosen": 0.872645378112793,
|
| 311 |
+
"logits/rejected": 0.8895421028137207,
|
| 312 |
+
"logps/chosen": -237.6038055419922,
|
| 313 |
+
"logps/rejected": -233.2046661376953,
|
| 314 |
+
"loss": 0.0023,
|
| 315 |
+
"rewards/accuracies": 0.581250011920929,
|
| 316 |
+
"rewards/chosen": -0.0021511532831937075,
|
| 317 |
+
"rewards/margins": 0.002740217139944434,
|
| 318 |
+
"rewards/rejected": -0.004891370423138142,
|
| 319 |
+
"step": 200
|
| 320 |
+
},
|
| 321 |
+
{
|
| 322 |
+
"epoch": 0.42,
|
| 323 |
+
"eval_logits/chosen": 0.8092045783996582,
|
| 324 |
+
"eval_logits/rejected": 0.9120355844497681,
|
| 325 |
+
"eval_logps/chosen": -260.39324951171875,
|
| 326 |
+
"eval_logps/rejected": -232.11520385742188,
|
| 327 |
+
"eval_loss": 0.0022759963758289814,
|
| 328 |
+
"eval_rewards/accuracies": 0.656000018119812,
|
| 329 |
+
"eval_rewards/chosen": -0.0014907378936186433,
|
| 330 |
+
"eval_rewards/margins": 0.005300111137330532,
|
| 331 |
+
"eval_rewards/rejected": -0.006790849845856428,
|
| 332 |
+
"eval_runtime": 327.2093,
|
| 333 |
+
"eval_samples_per_second": 6.112,
|
| 334 |
+
"eval_steps_per_second": 0.382,
|
| 335 |
+
"step": 200
|
| 336 |
+
},
|
| 337 |
+
{
|
| 338 |
+
"epoch": 0.44,
|
| 339 |
+
"learning_rate": 4.785350472409792e-06,
|
| 340 |
+
"logits/chosen": 0.9010735750198364,
|
| 341 |
+
"logits/rejected": 0.9061653017997742,
|
| 342 |
+
"logps/chosen": -232.94741821289062,
|
| 343 |
+
"logps/rejected": -228.1055145263672,
|
| 344 |
+
"loss": 0.0023,
|
| 345 |
+
"rewards/accuracies": 0.6875,
|
| 346 |
+
"rewards/chosen": -0.003469156799837947,
|
| 347 |
+
"rewards/margins": 0.003427647752687335,
|
| 348 |
+
"rewards/rejected": -0.006896805018186569,
|
| 349 |
+
"step": 210
|
| 350 |
+
},
|
| 351 |
+
{
|
| 352 |
+
"epoch": 0.46,
|
| 353 |
+
"learning_rate": 4.746717530629565e-06,
|
| 354 |
+
"logits/chosen": 0.8498672246932983,
|
| 355 |
+
"logits/rejected": 0.8839853405952454,
|
| 356 |
+
"logps/chosen": -259.6465148925781,
|
| 357 |
+
"logps/rejected": -238.8597869873047,
|
| 358 |
+
"loss": 0.0023,
|
| 359 |
+
"rewards/accuracies": 0.6499999761581421,
|
| 360 |
+
"rewards/chosen": -0.002919531427323818,
|
| 361 |
+
"rewards/margins": 0.00543471472337842,
|
| 362 |
+
"rewards/rejected": -0.00835424568504095,
|
| 363 |
+
"step": 220
|
| 364 |
+
},
|
| 365 |
+
{
|
| 366 |
+
"epoch": 0.48,
|
| 367 |
+
"learning_rate": 4.7050727958301505e-06,
|
| 368 |
+
"logits/chosen": 0.8881880044937134,
|
| 369 |
+
"logits/rejected": 0.8712642788887024,
|
| 370 |
+
"logps/chosen": -244.35507202148438,
|
| 371 |
+
"logps/rejected": -229.05136108398438,
|
| 372 |
+
"loss": 0.0022,
|
| 373 |
+
"rewards/accuracies": 0.637499988079071,
|
| 374 |
+
"rewards/chosen": -0.004195825196802616,
|
| 375 |
+
"rewards/margins": 0.004726833663880825,
|
| 376 |
+
"rewards/rejected": -0.008922659792006016,
|
| 377 |
+
"step": 230
|
| 378 |
+
},
|
| 379 |
+
{
|
| 380 |
+
"epoch": 0.5,
|
| 381 |
+
"learning_rate": 4.660472094042121e-06,
|
| 382 |
+
"logits/chosen": 0.8501306772232056,
|
| 383 |
+
"logits/rejected": 0.8583809733390808,
|
| 384 |
+
"logps/chosen": -281.69854736328125,
|
| 385 |
+
"logps/rejected": -234.6123504638672,
|
| 386 |
+
"loss": 0.0022,
|
| 387 |
+
"rewards/accuracies": 0.675000011920929,
|
| 388 |
+
"rewards/chosen": -0.004014792386442423,
|
| 389 |
+
"rewards/margins": 0.0071690999902784824,
|
| 390 |
+
"rewards/rejected": -0.011183892376720905,
|
| 391 |
+
"step": 240
|
| 392 |
+
},
|
| 393 |
+
{
|
| 394 |
+
"epoch": 0.52,
|
| 395 |
+
"learning_rate": 4.612975213859487e-06,
|
| 396 |
+
"logits/chosen": 0.8360323905944824,
|
| 397 |
+
"logits/rejected": 0.9075163006782532,
|
| 398 |
+
"logps/chosen": -268.0133361816406,
|
| 399 |
+
"logps/rejected": -241.88339233398438,
|
| 400 |
+
"loss": 0.0023,
|
| 401 |
+
"rewards/accuracies": 0.643750011920929,
|
| 402 |
+
"rewards/chosen": -0.0047610728070139885,
|
| 403 |
+
"rewards/margins": 0.0054484582506120205,
|
| 404 |
+
"rewards/rejected": -0.010209531523287296,
|
| 405 |
+
"step": 250
|
| 406 |
+
},
|
| 407 |
+
{
|
| 408 |
+
"epoch": 0.54,
|
| 409 |
+
"learning_rate": 4.5626458262912745e-06,
|
| 410 |
+
"logits/chosen": 0.8462463617324829,
|
| 411 |
+
"logits/rejected": 0.8862007260322571,
|
| 412 |
+
"logps/chosen": -274.3870544433594,
|
| 413 |
+
"logps/rejected": -260.1866149902344,
|
| 414 |
+
"loss": 0.0022,
|
| 415 |
+
"rewards/accuracies": 0.699999988079071,
|
| 416 |
+
"rewards/chosen": -0.002204451011493802,
|
| 417 |
+
"rewards/margins": 0.007608965039253235,
|
| 418 |
+
"rewards/rejected": -0.009813414886593819,
|
| 419 |
+
"step": 260
|
| 420 |
+
},
|
| 421 |
+
{
|
| 422 |
+
"epoch": 0.57,
|
| 423 |
+
"learning_rate": 4.509551399408598e-06,
|
| 424 |
+
"logits/chosen": 0.9279536008834839,
|
| 425 |
+
"logits/rejected": 0.9327741861343384,
|
| 426 |
+
"logps/chosen": -252.70175170898438,
|
| 427 |
+
"logps/rejected": -209.066650390625,
|
| 428 |
+
"loss": 0.0022,
|
| 429 |
+
"rewards/accuracies": 0.637499988079071,
|
| 430 |
+
"rewards/chosen": -0.004289106000214815,
|
| 431 |
+
"rewards/margins": 0.007459082640707493,
|
| 432 |
+
"rewards/rejected": -0.01174818817526102,
|
| 433 |
+
"step": 270
|
| 434 |
+
},
|
| 435 |
+
{
|
| 436 |
+
"epoch": 0.59,
|
| 437 |
+
"learning_rate": 4.453763107901676e-06,
|
| 438 |
+
"logits/chosen": 0.9238850474357605,
|
| 439 |
+
"logits/rejected": 0.8847238421440125,
|
| 440 |
+
"logps/chosen": -246.0789031982422,
|
| 441 |
+
"logps/rejected": -253.26327514648438,
|
| 442 |
+
"loss": 0.0023,
|
| 443 |
+
"rewards/accuracies": 0.5375000238418579,
|
| 444 |
+
"rewards/chosen": -0.007073036395013332,
|
| 445 |
+
"rewards/margins": 0.0031865164637565613,
|
| 446 |
+
"rewards/rejected": -0.010259552858769894,
|
| 447 |
+
"step": 280
|
| 448 |
+
},
|
| 449 |
+
{
|
| 450 |
+
"epoch": 0.61,
|
| 451 |
+
"learning_rate": 4.3953557376679856e-06,
|
| 452 |
+
"logits/chosen": 0.8482117652893066,
|
| 453 |
+
"logits/rejected": 0.8521090745925903,
|
| 454 |
+
"logps/chosen": -260.2223815917969,
|
| 455 |
+
"logps/rejected": -255.7244873046875,
|
| 456 |
+
"loss": 0.0022,
|
| 457 |
+
"rewards/accuracies": 0.5874999761581421,
|
| 458 |
+
"rewards/chosen": -0.006550622172653675,
|
| 459 |
+
"rewards/margins": 0.004290700424462557,
|
| 460 |
+
"rewards/rejected": -0.01084132306277752,
|
| 461 |
+
"step": 290
|
| 462 |
+
},
|
| 463 |
+
{
|
| 464 |
+
"epoch": 0.63,
|
| 465 |
+
"learning_rate": 4.33440758555951e-06,
|
| 466 |
+
"logits/chosen": 0.8257355690002441,
|
| 467 |
+
"logits/rejected": 0.9032806158065796,
|
| 468 |
+
"logps/chosen": -249.4975128173828,
|
| 469 |
+
"logps/rejected": -243.69088745117188,
|
| 470 |
+
"loss": 0.0022,
|
| 471 |
+
"rewards/accuracies": 0.6875,
|
| 472 |
+
"rewards/chosen": -0.004273009952157736,
|
| 473 |
+
"rewards/margins": 0.008262387476861477,
|
| 474 |
+
"rewards/rejected": -0.0125353978946805,
|
| 475 |
+
"step": 300
|
| 476 |
+
},
|
| 477 |
+
{
|
| 478 |
+
"epoch": 0.63,
|
| 479 |
+
"eval_logits/chosen": 0.7992061972618103,
|
| 480 |
+
"eval_logits/rejected": 0.9022300243377686,
|
| 481 |
+
"eval_logps/chosen": -260.91790771484375,
|
| 482 |
+
"eval_logps/rejected": -232.84466552734375,
|
| 483 |
+
"eval_loss": 0.002206910401582718,
|
| 484 |
+
"eval_rewards/accuracies": 0.6700000166893005,
|
| 485 |
+
"eval_rewards/chosen": -0.006737456191331148,
|
| 486 |
+
"eval_rewards/margins": 0.007348266430199146,
|
| 487 |
+
"eval_rewards/rejected": -0.014085723087191582,
|
| 488 |
+
"eval_runtime": 327.3645,
|
| 489 |
+
"eval_samples_per_second": 6.109,
|
| 490 |
+
"eval_steps_per_second": 0.382,
|
| 491 |
+
"step": 300
|
| 492 |
+
},
|
| 493 |
+
{
|
| 494 |
+
"epoch": 0.65,
|
| 495 |
+
"learning_rate": 4.2710003544234255e-06,
|
| 496 |
+
"logits/chosen": 0.851446270942688,
|
| 497 |
+
"logits/rejected": 0.8612054586410522,
|
| 498 |
+
"logps/chosen": -235.674072265625,
|
| 499 |
+
"logps/rejected": -225.989990234375,
|
| 500 |
+
"loss": 0.0022,
|
| 501 |
+
"rewards/accuracies": 0.637499988079071,
|
| 502 |
+
"rewards/chosen": -0.007650519721210003,
|
| 503 |
+
"rewards/margins": 0.006462027784436941,
|
| 504 |
+
"rewards/rejected": -0.014112548902630806,
|
| 505 |
+
"step": 310
|
| 506 |
+
},
|
| 507 |
+
{
|
| 508 |
+
"epoch": 0.67,
|
| 509 |
+
"learning_rate": 4.205219043576955e-06,
|
| 510 |
+
"logits/chosen": 0.859449565410614,
|
| 511 |
+
"logits/rejected": 0.8806228637695312,
|
| 512 |
+
"logps/chosen": -224.012451171875,
|
| 513 |
+
"logps/rejected": -215.5794677734375,
|
| 514 |
+
"loss": 0.0021,
|
| 515 |
+
"rewards/accuracies": 0.65625,
|
| 516 |
+
"rewards/chosen": -0.008152343332767487,
|
| 517 |
+
"rewards/margins": 0.006258256733417511,
|
| 518 |
+
"rewards/rejected": -0.014410600066184998,
|
| 519 |
+
"step": 320
|
| 520 |
+
},
|
| 521 |
+
{
|
| 522 |
+
"epoch": 0.69,
|
| 523 |
+
"learning_rate": 4.137151834863213e-06,
|
| 524 |
+
"logits/chosen": 0.8416824340820312,
|
| 525 |
+
"logits/rejected": 0.8180407285690308,
|
| 526 |
+
"logps/chosen": -253.04629516601562,
|
| 527 |
+
"logps/rejected": -258.2658386230469,
|
| 528 |
+
"loss": 0.0022,
|
| 529 |
+
"rewards/accuracies": 0.625,
|
| 530 |
+
"rewards/chosen": -0.007537019904702902,
|
| 531 |
+
"rewards/margins": 0.008035494945943356,
|
| 532 |
+
"rewards/rejected": -0.015572515316307545,
|
| 533 |
+
"step": 330
|
| 534 |
+
},
|
| 535 |
+
{
|
| 536 |
+
"epoch": 0.71,
|
| 537 |
+
"learning_rate": 4.066889974440757e-06,
|
| 538 |
+
"logits/chosen": 0.8714286088943481,
|
| 539 |
+
"logits/rejected": 0.870419979095459,
|
| 540 |
+
"logps/chosen": -266.2498474121094,
|
| 541 |
+
"logps/rejected": -248.62814331054688,
|
| 542 |
+
"loss": 0.0021,
|
| 543 |
+
"rewards/accuracies": 0.625,
|
| 544 |
+
"rewards/chosen": -0.006631535477936268,
|
| 545 |
+
"rewards/margins": 0.007086685858666897,
|
| 546 |
+
"rewards/rejected": -0.01371822226792574,
|
| 547 |
+
"step": 340
|
| 548 |
+
},
|
| 549 |
+
{
|
| 550 |
+
"epoch": 0.73,
|
| 551 |
+
"learning_rate": 3.994527650465352e-06,
|
| 552 |
+
"logits/chosen": 0.8660160899162292,
|
| 553 |
+
"logits/rejected": 0.8818486928939819,
|
| 554 |
+
"logps/chosen": -222.5798797607422,
|
| 555 |
+
"logps/rejected": -204.8966522216797,
|
| 556 |
+
"loss": 0.0022,
|
| 557 |
+
"rewards/accuracies": 0.625,
|
| 558 |
+
"rewards/chosen": -0.009565680287778378,
|
| 559 |
+
"rewards/margins": 0.006333778612315655,
|
| 560 |
+
"rewards/rejected": -0.015899458900094032,
|
| 561 |
+
"step": 350
|
| 562 |
+
},
|
| 563 |
+
{
|
| 564 |
+
"epoch": 0.75,
|
| 565 |
+
"learning_rate": 3.92016186682789e-06,
|
| 566 |
+
"logits/chosen": 0.7974532842636108,
|
| 567 |
+
"logits/rejected": 0.802233874797821,
|
| 568 |
+
"logps/chosen": -208.30282592773438,
|
| 569 |
+
"logps/rejected": -235.695556640625,
|
| 570 |
+
"loss": 0.0021,
|
| 571 |
+
"rewards/accuracies": 0.706250011920929,
|
| 572 |
+
"rewards/chosen": -0.008728450164198875,
|
| 573 |
+
"rewards/margins": 0.008383492939174175,
|
| 574 |
+
"rewards/rejected": -0.017111944034695625,
|
| 575 |
+
"step": 360
|
| 576 |
+
},
|
| 577 |
+
{
|
| 578 |
+
"epoch": 0.77,
|
| 579 |
+
"learning_rate": 3.843892313117724e-06,
|
| 580 |
+
"logits/chosen": 0.8825100064277649,
|
| 581 |
+
"logits/rejected": 0.9040519595146179,
|
| 582 |
+
"logps/chosen": -268.5409851074219,
|
| 583 |
+
"logps/rejected": -241.93862915039062,
|
| 584 |
+
"loss": 0.0022,
|
| 585 |
+
"rewards/accuracies": 0.6312500238418579,
|
| 586 |
+
"rewards/chosen": -0.008966553956270218,
|
| 587 |
+
"rewards/margins": 0.00720653822645545,
|
| 588 |
+
"rewards/rejected": -0.016173092648386955,
|
| 589 |
+
"step": 370
|
| 590 |
+
},
|
| 591 |
+
{
|
| 592 |
+
"epoch": 0.8,
|
| 593 |
+
"learning_rate": 3.7658212309857576e-06,
|
| 594 |
+
"logits/chosen": 0.8589727282524109,
|
| 595 |
+
"logits/rejected": 0.8763955235481262,
|
| 596 |
+
"logps/chosen": -238.5310516357422,
|
| 597 |
+
"logps/rejected": -214.51651000976562,
|
| 598 |
+
"loss": 0.0021,
|
| 599 |
+
"rewards/accuracies": 0.668749988079071,
|
| 600 |
+
"rewards/chosen": -0.008340245112776756,
|
| 601 |
+
"rewards/margins": 0.009361723437905312,
|
| 602 |
+
"rewards/rejected": -0.017701968550682068,
|
| 603 |
+
"step": 380
|
| 604 |
+
},
|
| 605 |
+
{
|
| 606 |
+
"epoch": 0.82,
|
| 607 |
+
"learning_rate": 3.686053277086401e-06,
|
| 608 |
+
"logits/chosen": 0.812663197517395,
|
| 609 |
+
"logits/rejected": 0.8946850895881653,
|
| 610 |
+
"logps/chosen": -261.9408874511719,
|
| 611 |
+
"logps/rejected": -236.6110382080078,
|
| 612 |
+
"loss": 0.0021,
|
| 613 |
+
"rewards/accuracies": 0.668749988079071,
|
| 614 |
+
"rewards/chosen": -0.008398517966270447,
|
| 615 |
+
"rewards/margins": 0.008632157929241657,
|
| 616 |
+
"rewards/rejected": -0.01703067496418953,
|
| 617 |
+
"step": 390
|
| 618 |
+
},
|
| 619 |
+
{
|
| 620 |
+
"epoch": 0.84,
|
| 621 |
+
"learning_rate": 3.604695382782159e-06,
|
| 622 |
+
"logits/chosen": 0.8278132677078247,
|
| 623 |
+
"logits/rejected": 0.8401390314102173,
|
| 624 |
+
"logps/chosen": -277.306396484375,
|
| 625 |
+
"logps/rejected": -252.70474243164062,
|
| 626 |
+
"loss": 0.0021,
|
| 627 |
+
"rewards/accuracies": 0.5874999761581421,
|
| 628 |
+
"rewards/chosen": -0.008408455178141594,
|
| 629 |
+
"rewards/margins": 0.007572343107312918,
|
| 630 |
+
"rewards/rejected": -0.0159807987511158,
|
| 631 |
+
"step": 400
|
| 632 |
+
},
|
| 633 |
+
{
|
| 634 |
+
"epoch": 0.84,
|
| 635 |
+
"eval_logits/chosen": 0.7884067893028259,
|
| 636 |
+
"eval_logits/rejected": 0.8913614153862,
|
| 637 |
+
"eval_logps/chosen": -261.16204833984375,
|
| 638 |
+
"eval_logps/rejected": -233.21571350097656,
|
| 639 |
+
"eval_loss": 0.0021709667053073645,
|
| 640 |
+
"eval_rewards/accuracies": 0.6639999747276306,
|
| 641 |
+
"eval_rewards/chosen": -0.009178930893540382,
|
| 642 |
+
"eval_rewards/margins": 0.008617207407951355,
|
| 643 |
+
"eval_rewards/rejected": -0.017796138301491737,
|
| 644 |
+
"eval_runtime": 327.7654,
|
| 645 |
+
"eval_samples_per_second": 6.102,
|
| 646 |
+
"eval_steps_per_second": 0.381,
|
| 647 |
+
"step": 400
|
| 648 |
+
},
|
| 649 |
+
{
|
| 650 |
+
"epoch": 0.86,
|
| 651 |
+
"learning_rate": 3.5218566107988872e-06,
|
| 652 |
+
"logits/chosen": 0.8511005640029907,
|
| 653 |
+
"logits/rejected": 0.8689740300178528,
|
| 654 |
+
"logps/chosen": -269.23675537109375,
|
| 655 |
+
"logps/rejected": -234.9521484375,
|
| 656 |
+
"loss": 0.0022,
|
| 657 |
+
"rewards/accuracies": 0.6875,
|
| 658 |
+
"rewards/chosen": -0.008150833658874035,
|
| 659 |
+
"rewards/margins": 0.009497146122157574,
|
| 660 |
+
"rewards/rejected": -0.01764797978103161,
|
| 661 |
+
"step": 410
|
| 662 |
+
},
|
| 663 |
+
{
|
| 664 |
+
"epoch": 0.88,
|
| 665 |
+
"learning_rate": 3.437648009023905e-06,
|
| 666 |
+
"logits/chosen": 0.796908974647522,
|
| 667 |
+
"logits/rejected": 0.8615992665290833,
|
| 668 |
+
"logps/chosen": -213.23800659179688,
|
| 669 |
+
"logps/rejected": -206.04647827148438,
|
| 670 |
+
"loss": 0.0022,
|
| 671 |
+
"rewards/accuracies": 0.59375,
|
| 672 |
+
"rewards/chosen": -0.009716427884995937,
|
| 673 |
+
"rewards/margins": 0.006270779762417078,
|
| 674 |
+
"rewards/rejected": -0.015987208113074303,
|
| 675 |
+
"step": 420
|
| 676 |
+
},
|
| 677 |
+
{
|
| 678 |
+
"epoch": 0.9,
|
| 679 |
+
"learning_rate": 3.352182461642929e-06,
|
| 680 |
+
"logits/chosen": 0.8233186602592468,
|
| 681 |
+
"logits/rejected": 0.8778685331344604,
|
| 682 |
+
"logps/chosen": -235.12954711914062,
|
| 683 |
+
"logps/rejected": -218.67416381835938,
|
| 684 |
+
"loss": 0.0021,
|
| 685 |
+
"rewards/accuracies": 0.737500011920929,
|
| 686 |
+
"rewards/chosen": -0.008281409740447998,
|
| 687 |
+
"rewards/margins": 0.008449633605778217,
|
| 688 |
+
"rewards/rejected": -0.01673104241490364,
|
| 689 |
+
"step": 430
|
| 690 |
+
},
|
| 691 |
+
{
|
| 692 |
+
"epoch": 0.92,
|
| 693 |
+
"learning_rate": 3.265574537815398e-06,
|
| 694 |
+
"logits/chosen": 0.8454925417900085,
|
| 695 |
+
"logits/rejected": 0.8766401410102844,
|
| 696 |
+
"logps/chosen": -280.82562255859375,
|
| 697 |
+
"logps/rejected": -242.23635864257812,
|
| 698 |
+
"loss": 0.0021,
|
| 699 |
+
"rewards/accuracies": 0.71875,
|
| 700 |
+
"rewards/chosen": -0.006689480505883694,
|
| 701 |
+
"rewards/margins": 0.010621527209877968,
|
| 702 |
+
"rewards/rejected": -0.017311008647084236,
|
| 703 |
+
"step": 440
|
| 704 |
+
},
|
| 705 |
+
{
|
| 706 |
+
"epoch": 0.94,
|
| 707 |
+
"learning_rate": 3.177940338091043e-06,
|
| 708 |
+
"logits/chosen": 0.8534622192382812,
|
| 709 |
+
"logits/rejected": 0.9055653810501099,
|
| 710 |
+
"logps/chosen": -252.76089477539062,
|
| 711 |
+
"logps/rejected": -221.62551879882812,
|
| 712 |
+
"loss": 0.0021,
|
| 713 |
+
"rewards/accuracies": 0.699999988079071,
|
| 714 |
+
"rewards/chosen": -0.007692619226872921,
|
| 715 |
+
"rewards/margins": 0.009279204532504082,
|
| 716 |
+
"rewards/rejected": -0.016971822828054428,
|
| 717 |
+
"step": 450
|
| 718 |
+
},
|
| 719 |
+
{
|
| 720 |
+
"epoch": 0.96,
|
| 721 |
+
"learning_rate": 3.089397338773569e-06,
|
| 722 |
+
"logits/chosen": 0.8498457074165344,
|
| 723 |
+
"logits/rejected": 0.8722270131111145,
|
| 724 |
+
"logps/chosen": -258.4722595214844,
|
| 725 |
+
"logps/rejected": -225.99447631835938,
|
| 726 |
+
"loss": 0.0021,
|
| 727 |
+
"rewards/accuracies": 0.668749988079071,
|
| 728 |
+
"rewards/chosen": -0.009471247904002666,
|
| 729 |
+
"rewards/margins": 0.008289327844977379,
|
| 730 |
+
"rewards/rejected": -0.01776057854294777,
|
| 731 |
+
"step": 460
|
| 732 |
+
},
|
| 733 |
+
{
|
| 734 |
+
"epoch": 0.98,
|
| 735 |
+
"learning_rate": 3.0000642344401115e-06,
|
| 736 |
+
"logits/chosen": 0.8206149935722351,
|
| 737 |
+
"logits/rejected": 0.8616384267807007,
|
| 738 |
+
"logps/chosen": -239.18264770507812,
|
| 739 |
+
"logps/rejected": -224.2777862548828,
|
| 740 |
+
"loss": 0.0021,
|
| 741 |
+
"rewards/accuracies": 0.581250011920929,
|
| 742 |
+
"rewards/chosen": -0.009907958097755909,
|
| 743 |
+
"rewards/margins": 0.007560922298580408,
|
| 744 |
+
"rewards/rejected": -0.017468880861997604,
|
| 745 |
+
"step": 470
|
| 746 |
+
},
|
| 747 |
+
{
|
| 748 |
+
"epoch": 1.0,
|
| 749 |
+
"learning_rate": 2.9100607788275547e-06,
|
| 750 |
+
"logits/chosen": 0.8556537628173828,
|
| 751 |
+
"logits/rejected": 0.9059454202651978,
|
| 752 |
+
"logps/chosen": -243.97933959960938,
|
| 753 |
+
"logps/rejected": -235.90817260742188,
|
| 754 |
+
"loss": 0.0022,
|
| 755 |
+
"rewards/accuracies": 0.6499999761581421,
|
| 756 |
+
"rewards/chosen": -0.009879620745778084,
|
| 757 |
+
"rewards/margins": 0.007370662875473499,
|
| 758 |
+
"rewards/rejected": -0.01725028082728386,
|
| 759 |
+
"step": 480
|
| 760 |
+
},
|
| 761 |
+
{
|
| 762 |
+
"epoch": 1.03,
|
| 763 |
+
"learning_rate": 2.8195076242990124e-06,
|
| 764 |
+
"logits/chosen": 0.837064266204834,
|
| 765 |
+
"logits/rejected": 0.8774217367172241,
|
| 766 |
+
"logps/chosen": -239.0427703857422,
|
| 767 |
+
"logps/rejected": -215.9508819580078,
|
| 768 |
+
"loss": 0.0021,
|
| 769 |
+
"rewards/accuracies": 0.675000011920929,
|
| 770 |
+
"rewards/chosen": -0.010479142889380455,
|
| 771 |
+
"rewards/margins": 0.008913186378777027,
|
| 772 |
+
"rewards/rejected": -0.019392330199480057,
|
| 773 |
+
"step": 490
|
| 774 |
+
},
|
| 775 |
+
{
|
| 776 |
+
"epoch": 1.05,
|
| 777 |
+
"learning_rate": 2.72852616010567e-06,
|
| 778 |
+
"logits/chosen": 0.8183882832527161,
|
| 779 |
+
"logits/rejected": 0.8531386256217957,
|
| 780 |
+
"logps/chosen": -253.97683715820312,
|
| 781 |
+
"logps/rejected": -234.94265747070312,
|
| 782 |
+
"loss": 0.0022,
|
| 783 |
+
"rewards/accuracies": 0.6937500238418579,
|
| 784 |
+
"rewards/chosen": -0.008641371503472328,
|
| 785 |
+
"rewards/margins": 0.009928500279784203,
|
| 786 |
+
"rewards/rejected": -0.01856987178325653,
|
| 787 |
+
"step": 500
|
| 788 |
+
},
|
| 789 |
+
{
|
| 790 |
+
"epoch": 1.05,
|
| 791 |
+
"eval_logits/chosen": 0.7821087837219238,
|
| 792 |
+
"eval_logits/rejected": 0.8852795362472534,
|
| 793 |
+
"eval_logps/chosen": -261.1851501464844,
|
| 794 |
+
"eval_logps/rejected": -233.3613739013672,
|
| 795 |
+
"eval_loss": 0.0021211737766861916,
|
| 796 |
+
"eval_rewards/accuracies": 0.7099999785423279,
|
| 797 |
+
"eval_rewards/chosen": -0.009409956634044647,
|
| 798 |
+
"eval_rewards/margins": 0.00984267145395279,
|
| 799 |
+
"eval_rewards/rejected": -0.019252628087997437,
|
| 800 |
+
"eval_runtime": 325.4617,
|
| 801 |
+
"eval_samples_per_second": 6.145,
|
| 802 |
+
"eval_steps_per_second": 0.384,
|
| 803 |
+
"step": 500
|
| 804 |
+
},
|
| 805 |
+
{
|
| 806 |
+
"epoch": 1.07,
|
| 807 |
+
"learning_rate": 2.637238349660819e-06,
|
| 808 |
+
"logits/chosen": 0.8250406384468079,
|
| 809 |
+
"logits/rejected": 0.9090532064437866,
|
| 810 |
+
"logps/chosen": -236.7153778076172,
|
| 811 |
+
"logps/rejected": -197.7652587890625,
|
| 812 |
+
"loss": 0.0021,
|
| 813 |
+
"rewards/accuracies": 0.637499988079071,
|
| 814 |
+
"rewards/chosen": -0.010482882149517536,
|
| 815 |
+
"rewards/margins": 0.00821294542402029,
|
| 816 |
+
"rewards/rejected": -0.018695827573537827,
|
| 817 |
+
"step": 510
|
| 818 |
+
},
|
| 819 |
+
{
|
| 820 |
+
"epoch": 1.09,
|
| 821 |
+
"learning_rate": 2.5457665670441937e-06,
|
| 822 |
+
"logits/chosen": 0.8841145634651184,
|
| 823 |
+
"logits/rejected": 0.891588568687439,
|
| 824 |
+
"logps/chosen": -250.2413330078125,
|
| 825 |
+
"logps/rejected": -226.20846557617188,
|
| 826 |
+
"loss": 0.0022,
|
| 827 |
+
"rewards/accuracies": 0.6499999761581421,
|
| 828 |
+
"rewards/chosen": -0.008791481144726276,
|
| 829 |
+
"rewards/margins": 0.008898518979549408,
|
| 830 |
+
"rewards/rejected": -0.01768999919295311,
|
| 831 |
+
"step": 520
|
| 832 |
+
},
|
| 833 |
+
{
|
| 834 |
+
"epoch": 1.11,
|
| 835 |
+
"learning_rate": 2.4542334329558075e-06,
|
| 836 |
+
"logits/chosen": 0.8093138933181763,
|
| 837 |
+
"logits/rejected": 0.8301302790641785,
|
| 838 |
+
"logps/chosen": -242.32766723632812,
|
| 839 |
+
"logps/rejected": -228.98446655273438,
|
| 840 |
+
"loss": 0.0021,
|
| 841 |
+
"rewards/accuracies": 0.7124999761581421,
|
| 842 |
+
"rewards/chosen": -0.008460971526801586,
|
| 843 |
+
"rewards/margins": 0.009179492481052876,
|
| 844 |
+
"rewards/rejected": -0.017640462145209312,
|
| 845 |
+
"step": 530
|
| 846 |
+
},
|
| 847 |
+
{
|
| 848 |
+
"epoch": 1.13,
|
| 849 |
+
"learning_rate": 2.3627616503391813e-06,
|
| 850 |
+
"logits/chosen": 0.8050084114074707,
|
| 851 |
+
"logits/rejected": 0.8223572969436646,
|
| 852 |
+
"logps/chosen": -259.1438903808594,
|
| 853 |
+
"logps/rejected": -216.0526885986328,
|
| 854 |
+
"loss": 0.0021,
|
| 855 |
+
"rewards/accuracies": 0.6312500238418579,
|
| 856 |
+
"rewards/chosen": -0.009418713860213757,
|
| 857 |
+
"rewards/margins": 0.0077890073880553246,
|
| 858 |
+
"rewards/rejected": -0.01720772124826908,
|
| 859 |
+
"step": 540
|
| 860 |
+
},
|
| 861 |
+
{
|
| 862 |
+
"epoch": 1.15,
|
| 863 |
+
"learning_rate": 2.271473839894331e-06,
|
| 864 |
+
"logits/chosen": 0.8079057931900024,
|
| 865 |
+
"logits/rejected": 0.8310044407844543,
|
| 866 |
+
"logps/chosen": -267.4736328125,
|
| 867 |
+
"logps/rejected": -249.3809356689453,
|
| 868 |
+
"loss": 0.0021,
|
| 869 |
+
"rewards/accuracies": 0.6499999761581421,
|
| 870 |
+
"rewards/chosen": -0.008644058369100094,
|
| 871 |
+
"rewards/margins": 0.008870486170053482,
|
| 872 |
+
"rewards/rejected": -0.01751454547047615,
|
| 873 |
+
"step": 550
|
| 874 |
+
},
|
| 875 |
+
{
|
| 876 |
+
"epoch": 1.17,
|
| 877 |
+
"learning_rate": 2.1804923757009885e-06,
|
| 878 |
+
"logits/chosen": 0.7828265428543091,
|
| 879 |
+
"logits/rejected": 0.810738205909729,
|
| 880 |
+
"logps/chosen": -251.98782348632812,
|
| 881 |
+
"logps/rejected": -226.2450408935547,
|
| 882 |
+
"loss": 0.0021,
|
| 883 |
+
"rewards/accuracies": 0.6937500238418579,
|
| 884 |
+
"rewards/chosen": -0.008495638146996498,
|
| 885 |
+
"rewards/margins": 0.009752650745213032,
|
| 886 |
+
"rewards/rejected": -0.018248289823532104,
|
| 887 |
+
"step": 560
|
| 888 |
+
},
|
| 889 |
+
{
|
| 890 |
+
"epoch": 1.19,
|
| 891 |
+
"learning_rate": 2.089939221172446e-06,
|
| 892 |
+
"logits/chosen": 0.8058856129646301,
|
| 893 |
+
"logits/rejected": 0.8096176385879517,
|
| 894 |
+
"logps/chosen": -270.29681396484375,
|
| 895 |
+
"logps/rejected": -230.7064666748047,
|
| 896 |
+
"loss": 0.0021,
|
| 897 |
+
"rewards/accuracies": 0.6937500238418579,
|
| 898 |
+
"rewards/chosen": -0.010026035830378532,
|
| 899 |
+
"rewards/margins": 0.009994433261454105,
|
| 900 |
+
"rewards/rejected": -0.020020468160510063,
|
| 901 |
+
"step": 570
|
| 902 |
+
},
|
| 903 |
+
{
|
| 904 |
+
"epoch": 1.21,
|
| 905 |
+
"learning_rate": 1.9999357655598894e-06,
|
| 906 |
+
"logits/chosen": 0.8202878832817078,
|
| 907 |
+
"logits/rejected": 0.7905790209770203,
|
| 908 |
+
"logps/chosen": -240.6394500732422,
|
| 909 |
+
"logps/rejected": -227.01150512695312,
|
| 910 |
+
"loss": 0.0022,
|
| 911 |
+
"rewards/accuracies": 0.6499999761581421,
|
| 912 |
+
"rewards/chosen": -0.007834648713469505,
|
| 913 |
+
"rewards/margins": 0.007603611797094345,
|
| 914 |
+
"rewards/rejected": -0.01543826051056385,
|
| 915 |
+
"step": 580
|
| 916 |
+
},
|
| 917 |
+
{
|
| 918 |
+
"epoch": 1.23,
|
| 919 |
+
"learning_rate": 1.9106026612264316e-06,
|
| 920 |
+
"logits/chosen": 0.8630008697509766,
|
| 921 |
+
"logits/rejected": 0.9225195050239563,
|
| 922 |
+
"logps/chosen": -218.4501953125,
|
| 923 |
+
"logps/rejected": -214.31399536132812,
|
| 924 |
+
"loss": 0.0021,
|
| 925 |
+
"rewards/accuracies": 0.637499988079071,
|
| 926 |
+
"rewards/chosen": -0.008840398862957954,
|
| 927 |
+
"rewards/margins": 0.009108386933803558,
|
| 928 |
+
"rewards/rejected": -0.017948785796761513,
|
| 929 |
+
"step": 590
|
| 930 |
+
},
|
| 931 |
+
{
|
| 932 |
+
"epoch": 1.26,
|
| 933 |
+
"learning_rate": 1.8220596619089576e-06,
|
| 934 |
+
"logits/chosen": 0.8304165601730347,
|
| 935 |
+
"logits/rejected": 0.7959300875663757,
|
| 936 |
+
"logps/chosen": -262.4768981933594,
|
| 937 |
+
"logps/rejected": -230.4768524169922,
|
| 938 |
+
"loss": 0.002,
|
| 939 |
+
"rewards/accuracies": 0.6937500238418579,
|
| 940 |
+
"rewards/chosen": -0.008658383972942829,
|
| 941 |
+
"rewards/margins": 0.008863124065101147,
|
| 942 |
+
"rewards/rejected": -0.017521508038043976,
|
| 943 |
+
"step": 600
|
| 944 |
+
},
|
| 945 |
+
{
|
| 946 |
+
"epoch": 1.26,
|
| 947 |
+
"eval_logits/chosen": 0.7814888954162598,
|
| 948 |
+
"eval_logits/rejected": 0.8839987516403198,
|
| 949 |
+
"eval_logps/chosen": -261.1206970214844,
|
| 950 |
+
"eval_logps/rejected": -233.28433227539062,
|
| 951 |
+
"eval_loss": 0.002118554199114442,
|
| 952 |
+
"eval_rewards/accuracies": 0.6940000057220459,
|
| 953 |
+
"eval_rewards/chosen": -0.008765296079218388,
|
| 954 |
+
"eval_rewards/margins": 0.009716734290122986,
|
| 955 |
+
"eval_rewards/rejected": -0.0184820294380188,
|
| 956 |
+
"eval_runtime": 325.1283,
|
| 957 |
+
"eval_samples_per_second": 6.151,
|
| 958 |
+
"eval_steps_per_second": 0.384,
|
| 959 |
+
"step": 600
|
| 960 |
+
},
|
| 961 |
+
{
|
| 962 |
+
"epoch": 1.28,
|
| 963 |
+
"learning_rate": 1.7344254621846018e-06,
|
| 964 |
+
"logits/chosen": 0.7952737212181091,
|
| 965 |
+
"logits/rejected": 0.8380396962165833,
|
| 966 |
+
"logps/chosen": -264.8700256347656,
|
| 967 |
+
"logps/rejected": -251.73214721679688,
|
| 968 |
+
"loss": 0.0019,
|
| 969 |
+
"rewards/accuracies": 0.75,
|
| 970 |
+
"rewards/chosen": -0.007588304579257965,
|
| 971 |
+
"rewards/margins": 0.011967618018388748,
|
| 972 |
+
"rewards/rejected": -0.019555922597646713,
|
| 973 |
+
"step": 610
|
| 974 |
+
},
|
| 975 |
+
{
|
| 976 |
+
"epoch": 1.3,
|
| 977 |
+
"learning_rate": 1.647817538357072e-06,
|
| 978 |
+
"logits/chosen": 0.8249796032905579,
|
| 979 |
+
"logits/rejected": 0.8937106132507324,
|
| 980 |
+
"logps/chosen": -247.3903350830078,
|
| 981 |
+
"logps/rejected": -229.36093139648438,
|
| 982 |
+
"loss": 0.0022,
|
| 983 |
+
"rewards/accuracies": 0.612500011920929,
|
| 984 |
+
"rewards/chosen": -0.010081231594085693,
|
| 985 |
+
"rewards/margins": 0.006175318732857704,
|
| 986 |
+
"rewards/rejected": -0.016256550326943398,
|
| 987 |
+
"step": 620
|
| 988 |
+
},
|
| 989 |
+
{
|
| 990 |
+
"epoch": 1.32,
|
| 991 |
+
"learning_rate": 1.5623519909760953e-06,
|
| 992 |
+
"logits/chosen": 0.8032970428466797,
|
| 993 |
+
"logits/rejected": 0.8601589202880859,
|
| 994 |
+
"logps/chosen": -242.910888671875,
|
| 995 |
+
"logps/rejected": -238.8945770263672,
|
| 996 |
+
"loss": 0.0021,
|
| 997 |
+
"rewards/accuracies": 0.668749988079071,
|
| 998 |
+
"rewards/chosen": -0.010011060163378716,
|
| 999 |
+
"rewards/margins": 0.007496376521885395,
|
| 1000 |
+
"rewards/rejected": -0.017507437616586685,
|
| 1001 |
+
"step": 630
|
| 1002 |
+
},
|
| 1003 |
+
{
|
| 1004 |
+
"epoch": 1.34,
|
| 1005 |
+
"learning_rate": 1.4781433892011132e-06,
|
| 1006 |
+
"logits/chosen": 0.8834539651870728,
|
| 1007 |
+
"logits/rejected": 0.8997892141342163,
|
| 1008 |
+
"logps/chosen": -242.17578125,
|
| 1009 |
+
"logps/rejected": -259.06671142578125,
|
| 1010 |
+
"loss": 0.0021,
|
| 1011 |
+
"rewards/accuracies": 0.668749988079071,
|
| 1012 |
+
"rewards/chosen": -0.008281581103801727,
|
| 1013 |
+
"rewards/margins": 0.008412448689341545,
|
| 1014 |
+
"rewards/rejected": -0.016694029793143272,
|
| 1015 |
+
"step": 640
|
| 1016 |
+
},
|
| 1017 |
+
{
|
| 1018 |
+
"epoch": 1.36,
|
| 1019 |
+
"learning_rate": 1.3953046172178413e-06,
|
| 1020 |
+
"logits/chosen": 0.8225077390670776,
|
| 1021 |
+
"logits/rejected": 0.8323785662651062,
|
| 1022 |
+
"logps/chosen": -253.43905639648438,
|
| 1023 |
+
"logps/rejected": -245.92721557617188,
|
| 1024 |
+
"loss": 0.0021,
|
| 1025 |
+
"rewards/accuracies": 0.6875,
|
| 1026 |
+
"rewards/chosen": -0.010358546860516071,
|
| 1027 |
+
"rewards/margins": 0.009153308346867561,
|
| 1028 |
+
"rewards/rejected": -0.019511854276061058,
|
| 1029 |
+
"step": 650
|
| 1030 |
+
},
|
| 1031 |
+
{
|
| 1032 |
+
"epoch": 1.38,
|
| 1033 |
+
"learning_rate": 1.3139467229135999e-06,
|
| 1034 |
+
"logits/chosen": 0.7964144945144653,
|
| 1035 |
+
"logits/rejected": 0.8561304807662964,
|
| 1036 |
+
"logps/chosen": -220.56430053710938,
|
| 1037 |
+
"logps/rejected": -246.96731567382812,
|
| 1038 |
+
"loss": 0.0022,
|
| 1039 |
+
"rewards/accuracies": 0.675000011920929,
|
| 1040 |
+
"rewards/chosen": -0.008580346591770649,
|
| 1041 |
+
"rewards/margins": 0.00769965723156929,
|
| 1042 |
+
"rewards/rejected": -0.016280004754662514,
|
| 1043 |
+
"step": 660
|
| 1044 |
+
},
|
| 1045 |
+
{
|
| 1046 |
+
"epoch": 1.4,
|
| 1047 |
+
"learning_rate": 1.2341787690142436e-06,
|
| 1048 |
+
"logits/chosen": 0.7782607078552246,
|
| 1049 |
+
"logits/rejected": 0.8575235605239868,
|
| 1050 |
+
"logps/chosen": -300.5128173828125,
|
| 1051 |
+
"logps/rejected": -240.5801239013672,
|
| 1052 |
+
"loss": 0.0021,
|
| 1053 |
+
"rewards/accuracies": 0.737500011920929,
|
| 1054 |
+
"rewards/chosen": -0.007187344133853912,
|
| 1055 |
+
"rewards/margins": 0.01008409820497036,
|
| 1056 |
+
"rewards/rejected": -0.017271442338824272,
|
| 1057 |
+
"step": 670
|
| 1058 |
+
},
|
| 1059 |
+
{
|
| 1060 |
+
"epoch": 1.42,
|
| 1061 |
+
"learning_rate": 1.1561076868822756e-06,
|
| 1062 |
+
"logits/chosen": 0.8010842204093933,
|
| 1063 |
+
"logits/rejected": 0.8400290608406067,
|
| 1064 |
+
"logps/chosen": -243.8752899169922,
|
| 1065 |
+
"logps/rejected": -216.2671661376953,
|
| 1066 |
+
"loss": 0.002,
|
| 1067 |
+
"rewards/accuracies": 0.6875,
|
| 1068 |
+
"rewards/chosen": -0.007206754293292761,
|
| 1069 |
+
"rewards/margins": 0.009658637456595898,
|
| 1070 |
+
"rewards/rejected": -0.01686539314687252,
|
| 1071 |
+
"step": 680
|
| 1072 |
+
},
|
| 1073 |
+
{
|
| 1074 |
+
"epoch": 1.44,
|
| 1075 |
+
"learning_rate": 1.079838133172111e-06,
|
| 1076 |
+
"logits/chosen": 0.8610042333602905,
|
| 1077 |
+
"logits/rejected": 0.8555776476860046,
|
| 1078 |
+
"logps/chosen": -258.38946533203125,
|
| 1079 |
+
"logps/rejected": -238.4031982421875,
|
| 1080 |
+
"loss": 0.0021,
|
| 1081 |
+
"rewards/accuracies": 0.668749988079071,
|
| 1082 |
+
"rewards/chosen": -0.008076001890003681,
|
| 1083 |
+
"rewards/margins": 0.0069586001336574554,
|
| 1084 |
+
"rewards/rejected": -0.015034601092338562,
|
| 1085 |
+
"step": 690
|
| 1086 |
+
},
|
| 1087 |
+
{
|
| 1088 |
+
"epoch": 1.47,
|
| 1089 |
+
"learning_rate": 1.0054723495346484e-06,
|
| 1090 |
+
"logits/chosen": 0.7388730049133301,
|
| 1091 |
+
"logits/rejected": 0.8165189027786255,
|
| 1092 |
+
"logps/chosen": -259.45965576171875,
|
| 1093 |
+
"logps/rejected": -227.28750610351562,
|
| 1094 |
+
"loss": 0.0021,
|
| 1095 |
+
"rewards/accuracies": 0.668749988079071,
|
| 1096 |
+
"rewards/chosen": -0.00880814902484417,
|
| 1097 |
+
"rewards/margins": 0.007600386627018452,
|
| 1098 |
+
"rewards/rejected": -0.016408536583185196,
|
| 1099 |
+
"step": 700
|
| 1100 |
+
},
|
| 1101 |
+
{
|
| 1102 |
+
"epoch": 1.47,
|
| 1103 |
+
"eval_logits/chosen": 0.7790006995201111,
|
| 1104 |
+
"eval_logits/rejected": 0.8815683722496033,
|
| 1105 |
+
"eval_logps/chosen": -261.0787658691406,
|
| 1106 |
+
"eval_logps/rejected": -233.25596618652344,
|
| 1107 |
+
"eval_loss": 0.0021041170693933964,
|
| 1108 |
+
"eval_rewards/accuracies": 0.699999988079071,
|
| 1109 |
+
"eval_rewards/chosen": -0.00834597460925579,
|
| 1110 |
+
"eval_rewards/margins": 0.009852716699242592,
|
| 1111 |
+
"eval_rewards/rejected": -0.018198693171143532,
|
| 1112 |
+
"eval_runtime": 325.1201,
|
| 1113 |
+
"eval_samples_per_second": 6.152,
|
| 1114 |
+
"eval_steps_per_second": 0.384,
|
| 1115 |
+
"step": 700
|
| 1116 |
+
},
|
| 1117 |
+
{
|
| 1118 |
+
"epoch": 1.49,
|
| 1119 |
+
"learning_rate": 9.331100255592437e-07,
|
| 1120 |
+
"logits/chosen": 0.8419380187988281,
|
| 1121 |
+
"logits/rejected": 0.8946343660354614,
|
| 1122 |
+
"logps/chosen": -279.4914245605469,
|
| 1123 |
+
"logps/rejected": -229.0105438232422,
|
| 1124 |
+
"loss": 0.0021,
|
| 1125 |
+
"rewards/accuracies": 0.6812499761581421,
|
| 1126 |
+
"rewards/chosen": -0.0075461543165147305,
|
| 1127 |
+
"rewards/margins": 0.010306203737854958,
|
| 1128 |
+
"rewards/rejected": -0.017852356657385826,
|
| 1129 |
+
"step": 710
|
| 1130 |
+
},
|
| 1131 |
+
{
|
| 1132 |
+
"epoch": 1.51,
|
| 1133 |
+
"learning_rate": 8.628481651367876e-07,
|
| 1134 |
+
"logits/chosen": 0.8257007598876953,
|
| 1135 |
+
"logits/rejected": 0.8564590215682983,
|
| 1136 |
+
"logps/chosen": -264.4073486328125,
|
| 1137 |
+
"logps/rejected": -244.4358673095703,
|
| 1138 |
+
"loss": 0.002,
|
| 1139 |
+
"rewards/accuracies": 0.65625,
|
| 1140 |
+
"rewards/chosen": -0.007830760441720486,
|
| 1141 |
+
"rewards/margins": 0.009762524627149105,
|
| 1142 |
+
"rewards/rejected": -0.01759328506886959,
|
| 1143 |
+
"step": 720
|
| 1144 |
+
},
|
| 1145 |
+
{
|
| 1146 |
+
"epoch": 1.53,
|
| 1147 |
+
"learning_rate": 7.947809564230446e-07,
|
| 1148 |
+
"logits/chosen": 0.8498528599739075,
|
| 1149 |
+
"logits/rejected": 0.8689123392105103,
|
| 1150 |
+
"logps/chosen": -265.91986083984375,
|
| 1151 |
+
"logps/rejected": -252.3328094482422,
|
| 1152 |
+
"loss": 0.002,
|
| 1153 |
+
"rewards/accuracies": 0.7749999761581421,
|
| 1154 |
+
"rewards/chosen": -0.007306605577468872,
|
| 1155 |
+
"rewards/margins": 0.00969378836452961,
|
| 1156 |
+
"rewards/rejected": -0.01700039580464363,
|
| 1157 |
+
"step": 730
|
| 1158 |
+
},
|
| 1159 |
+
{
|
| 1160 |
+
"epoch": 1.55,
|
| 1161 |
+
"learning_rate": 7.289996455765749e-07,
|
| 1162 |
+
"logits/chosen": 0.8058244585990906,
|
| 1163 |
+
"logits/rejected": 0.8203352093696594,
|
| 1164 |
+
"logps/chosen": -276.05224609375,
|
| 1165 |
+
"logps/rejected": -252.39712524414062,
|
| 1166 |
+
"loss": 0.0021,
|
| 1167 |
+
"rewards/accuracies": 0.6499999761581421,
|
| 1168 |
+
"rewards/chosen": -0.00796983577311039,
|
| 1169 |
+
"rewards/margins": 0.007479208521544933,
|
| 1170 |
+
"rewards/rejected": -0.015449045225977898,
|
| 1171 |
+
"step": 740
|
| 1172 |
+
},
|
| 1173 |
+
{
|
| 1174 |
+
"epoch": 1.57,
|
| 1175 |
+
"learning_rate": 6.655924144404907e-07,
|
| 1176 |
+
"logits/chosen": 0.7636196613311768,
|
| 1177 |
+
"logits/rejected": 0.8203707933425903,
|
| 1178 |
+
"logps/chosen": -260.9599304199219,
|
| 1179 |
+
"logps/rejected": -236.13442993164062,
|
| 1180 |
+
"loss": 0.0021,
|
| 1181 |
+
"rewards/accuracies": 0.7250000238418579,
|
| 1182 |
+
"rewards/chosen": -0.0055490517988801,
|
| 1183 |
+
"rewards/margins": 0.010982049629092216,
|
| 1184 |
+
"rewards/rejected": -0.01653110235929489,
|
| 1185 |
+
"step": 750
|
| 1186 |
+
},
|
| 1187 |
+
{
|
| 1188 |
+
"epoch": 1.59,
|
| 1189 |
+
"learning_rate": 6.046442623320145e-07,
|
| 1190 |
+
"logits/chosen": 0.8590337634086609,
|
| 1191 |
+
"logits/rejected": 0.889240562915802,
|
| 1192 |
+
"logps/chosen": -256.20843505859375,
|
| 1193 |
+
"logps/rejected": -218.8590087890625,
|
| 1194 |
+
"loss": 0.0021,
|
| 1195 |
+
"rewards/accuracies": 0.612500011920929,
|
| 1196 |
+
"rewards/chosen": -0.009814934805035591,
|
| 1197 |
+
"rewards/margins": 0.007059067487716675,
|
| 1198 |
+
"rewards/rejected": -0.016874000430107117,
|
| 1199 |
+
"step": 760
|
| 1200 |
+
},
|
| 1201 |
+
{
|
| 1202 |
+
"epoch": 1.61,
|
| 1203 |
+
"learning_rate": 5.462368920983249e-07,
|
| 1204 |
+
"logits/chosen": 0.8207941055297852,
|
| 1205 |
+
"logits/rejected": 0.8683902025222778,
|
| 1206 |
+
"logps/chosen": -241.982666015625,
|
| 1207 |
+
"logps/rejected": -217.79141235351562,
|
| 1208 |
+
"loss": 0.0022,
|
| 1209 |
+
"rewards/accuracies": 0.6499999761581421,
|
| 1210 |
+
"rewards/chosen": -0.009196323342621326,
|
| 1211 |
+
"rewards/margins": 0.006100159604102373,
|
| 1212 |
+
"rewards/rejected": -0.015296483412384987,
|
| 1213 |
+
"step": 770
|
| 1214 |
+
},
|
| 1215 |
+
{
|
| 1216 |
+
"epoch": 1.63,
|
| 1217 |
+
"learning_rate": 4.904486005914027e-07,
|
| 1218 |
+
"logits/chosen": 0.8281243443489075,
|
| 1219 |
+
"logits/rejected": 0.8692743182182312,
|
| 1220 |
+
"logps/chosen": -284.8575744628906,
|
| 1221 |
+
"logps/rejected": -258.57464599609375,
|
| 1222 |
+
"loss": 0.002,
|
| 1223 |
+
"rewards/accuracies": 0.675000011920929,
|
| 1224 |
+
"rewards/chosen": -0.0074394033290445805,
|
| 1225 |
+
"rewards/margins": 0.010337688960134983,
|
| 1226 |
+
"rewards/rejected": -0.01777709275484085,
|
| 1227 |
+
"step": 780
|
| 1228 |
+
},
|
| 1229 |
+
{
|
| 1230 |
+
"epoch": 1.65,
|
| 1231 |
+
"learning_rate": 4.373541737087264e-07,
|
| 1232 |
+
"logits/chosen": 0.8097678422927856,
|
| 1233 |
+
"logits/rejected": 0.9018150568008423,
|
| 1234 |
+
"logps/chosen": -265.8307189941406,
|
| 1235 |
+
"logps/rejected": -224.81103515625,
|
| 1236 |
+
"loss": 0.0022,
|
| 1237 |
+
"rewards/accuracies": 0.668749988079071,
|
| 1238 |
+
"rewards/chosen": -0.007008875720202923,
|
| 1239 |
+
"rewards/margins": 0.009901894256472588,
|
| 1240 |
+
"rewards/rejected": -0.016910770907998085,
|
| 1241 |
+
"step": 790
|
| 1242 |
+
},
|
| 1243 |
+
{
|
| 1244 |
+
"epoch": 1.67,
|
| 1245 |
+
"learning_rate": 3.8702478614051353e-07,
|
| 1246 |
+
"logits/chosen": 0.7832424640655518,
|
| 1247 |
+
"logits/rejected": 0.7852426767349243,
|
| 1248 |
+
"logps/chosen": -238.4669952392578,
|
| 1249 |
+
"logps/rejected": -229.850830078125,
|
| 1250 |
+
"loss": 0.0021,
|
| 1251 |
+
"rewards/accuracies": 0.737500011920929,
|
| 1252 |
+
"rewards/chosen": -0.007307013962417841,
|
| 1253 |
+
"rewards/margins": 0.011516671627759933,
|
| 1254 |
+
"rewards/rejected": -0.018823683261871338,
|
| 1255 |
+
"step": 800
|
| 1256 |
+
},
|
| 1257 |
+
{
|
| 1258 |
+
"epoch": 1.67,
|
| 1259 |
+
"eval_logits/chosen": 0.7780749797821045,
|
| 1260 |
+
"eval_logits/rejected": 0.8810694217681885,
|
| 1261 |
+
"eval_logps/chosen": -261.0643310546875,
|
| 1262 |
+
"eval_logps/rejected": -233.27398681640625,
|
| 1263 |
+
"eval_loss": 0.002105255611240864,
|
| 1264 |
+
"eval_rewards/accuracies": 0.6940000057220459,
|
| 1265 |
+
"eval_rewards/chosen": -0.008201568387448788,
|
| 1266 |
+
"eval_rewards/margins": 0.010177312418818474,
|
| 1267 |
+
"eval_rewards/rejected": -0.018378881737589836,
|
| 1268 |
+
"eval_runtime": 324.9688,
|
| 1269 |
+
"eval_samples_per_second": 6.154,
|
| 1270 |
+
"eval_steps_per_second": 0.385,
|
| 1271 |
+
"step": 800
|
| 1272 |
+
},
|
| 1273 |
+
{
|
| 1274 |
+
"epoch": 1.7,
|
| 1275 |
+
"learning_rate": 3.3952790595787986e-07,
|
| 1276 |
+
"logits/chosen": 0.9047285914421082,
|
| 1277 |
+
"logits/rejected": 0.8632117509841919,
|
| 1278 |
+
"logps/chosen": -249.9727020263672,
|
| 1279 |
+
"logps/rejected": -233.0714569091797,
|
| 1280 |
+
"loss": 0.0021,
|
| 1281 |
+
"rewards/accuracies": 0.606249988079071,
|
| 1282 |
+
"rewards/chosen": -0.008487801998853683,
|
| 1283 |
+
"rewards/margins": 0.007260690443217754,
|
| 1284 |
+
"rewards/rejected": -0.015748491510748863,
|
| 1285 |
+
"step": 810
|
| 1286 |
+
},
|
| 1287 |
+
{
|
| 1288 |
+
"epoch": 1.72,
|
| 1289 |
+
"learning_rate": 2.9492720416985004e-07,
|
| 1290 |
+
"logits/chosen": 0.745966374874115,
|
| 1291 |
+
"logits/rejected": 0.7877084016799927,
|
| 1292 |
+
"logps/chosen": -275.1504211425781,
|
| 1293 |
+
"logps/rejected": -252.06570434570312,
|
| 1294 |
+
"loss": 0.0021,
|
| 1295 |
+
"rewards/accuracies": 0.65625,
|
| 1296 |
+
"rewards/chosen": -0.006994744297116995,
|
| 1297 |
+
"rewards/margins": 0.008207054808735847,
|
| 1298 |
+
"rewards/rejected": -0.01520179957151413,
|
| 1299 |
+
"step": 820
|
| 1300 |
+
},
|
| 1301 |
+
{
|
| 1302 |
+
"epoch": 1.74,
|
| 1303 |
+
"learning_rate": 2.5328246937043526e-07,
|
| 1304 |
+
"logits/chosen": 0.8438940048217773,
|
| 1305 |
+
"logits/rejected": 0.8585928082466125,
|
| 1306 |
+
"logps/chosen": -237.1392059326172,
|
| 1307 |
+
"logps/rejected": -217.2599334716797,
|
| 1308 |
+
"loss": 0.0021,
|
| 1309 |
+
"rewards/accuracies": 0.668749988079071,
|
| 1310 |
+
"rewards/chosen": -0.009680895134806633,
|
| 1311 |
+
"rewards/margins": 0.008195875212550163,
|
| 1312 |
+
"rewards/rejected": -0.017876770347356796,
|
| 1313 |
+
"step": 830
|
| 1314 |
+
},
|
| 1315 |
+
{
|
| 1316 |
+
"epoch": 1.76,
|
| 1317 |
+
"learning_rate": 2.1464952759020857e-07,
|
| 1318 |
+
"logits/chosen": 0.835778534412384,
|
| 1319 |
+
"logits/rejected": 0.8605103492736816,
|
| 1320 |
+
"logps/chosen": -250.32455444335938,
|
| 1321 |
+
"logps/rejected": -247.2939453125,
|
| 1322 |
+
"loss": 0.0021,
|
| 1323 |
+
"rewards/accuracies": 0.6875,
|
| 1324 |
+
"rewards/chosen": -0.006872941739857197,
|
| 1325 |
+
"rewards/margins": 0.00862927082926035,
|
| 1326 |
+
"rewards/rejected": -0.015502211637794971,
|
| 1327 |
+
"step": 840
|
| 1328 |
+
},
|
| 1329 |
+
{
|
| 1330 |
+
"epoch": 1.78,
|
| 1331 |
+
"learning_rate": 1.790801674598186e-07,
|
| 1332 |
+
"logits/chosen": 0.8059646487236023,
|
| 1333 |
+
"logits/rejected": 0.8204299807548523,
|
| 1334 |
+
"logps/chosen": -269.0479736328125,
|
| 1335 |
+
"logps/rejected": -236.54345703125,
|
| 1336 |
+
"loss": 0.0021,
|
| 1337 |
+
"rewards/accuracies": 0.75,
|
| 1338 |
+
"rewards/chosen": -0.007251231465488672,
|
| 1339 |
+
"rewards/margins": 0.010516216047108173,
|
| 1340 |
+
"rewards/rejected": -0.017767447978258133,
|
| 1341 |
+
"step": 850
|
| 1342 |
+
},
|
| 1343 |
+
{
|
| 1344 |
+
"epoch": 1.8,
|
| 1345 |
+
"learning_rate": 1.4662207078575685e-07,
|
| 1346 |
+
"logits/chosen": 0.8005737066268921,
|
| 1347 |
+
"logits/rejected": 0.865730881690979,
|
| 1348 |
+
"logps/chosen": -232.56271362304688,
|
| 1349 |
+
"logps/rejected": -240.54428100585938,
|
| 1350 |
+
"loss": 0.0021,
|
| 1351 |
+
"rewards/accuracies": 0.71875,
|
| 1352 |
+
"rewards/chosen": -0.008310760371387005,
|
| 1353 |
+
"rewards/margins": 0.010277243331074715,
|
| 1354 |
+
"rewards/rejected": -0.018588004633784294,
|
| 1355 |
+
"step": 860
|
| 1356 |
+
},
|
| 1357 |
+
{
|
| 1358 |
+
"epoch": 1.82,
|
| 1359 |
+
"learning_rate": 1.1731874863145143e-07,
|
| 1360 |
+
"logits/chosen": 0.8402239680290222,
|
| 1361 |
+
"logits/rejected": 0.8874310255050659,
|
| 1362 |
+
"logps/chosen": -262.51519775390625,
|
| 1363 |
+
"logps/rejected": -224.810302734375,
|
| 1364 |
+
"loss": 0.002,
|
| 1365 |
+
"rewards/accuracies": 0.7124999761581421,
|
| 1366 |
+
"rewards/chosen": -0.006995867937803268,
|
| 1367 |
+
"rewards/margins": 0.009674609638750553,
|
| 1368 |
+
"rewards/rejected": -0.016670476645231247,
|
| 1369 |
+
"step": 870
|
| 1370 |
+
},
|
| 1371 |
+
{
|
| 1372 |
+
"epoch": 1.84,
|
| 1373 |
+
"learning_rate": 9.120948298936422e-08,
|
| 1374 |
+
"logits/chosen": 0.8549894094467163,
|
| 1375 |
+
"logits/rejected": 0.8683481216430664,
|
| 1376 |
+
"logps/chosen": -250.1100311279297,
|
| 1377 |
+
"logps/rejected": -225.083740234375,
|
| 1378 |
+
"loss": 0.0019,
|
| 1379 |
+
"rewards/accuracies": 0.6499999761581421,
|
| 1380 |
+
"rewards/chosen": -0.007406042423099279,
|
| 1381 |
+
"rewards/margins": 0.009078353643417358,
|
| 1382 |
+
"rewards/rejected": -0.016484394669532776,
|
| 1383 |
+
"step": 880
|
| 1384 |
+
},
|
| 1385 |
+
{
|
| 1386 |
+
"epoch": 1.86,
|
| 1387 |
+
"learning_rate": 6.832927412229017e-08,
|
| 1388 |
+
"logits/chosen": 0.785990834236145,
|
| 1389 |
+
"logits/rejected": 0.9018427729606628,
|
| 1390 |
+
"logps/chosen": -262.3821716308594,
|
| 1391 |
+
"logps/rejected": -228.41357421875,
|
| 1392 |
+
"loss": 0.002,
|
| 1393 |
+
"rewards/accuracies": 0.699999988079071,
|
| 1394 |
+
"rewards/chosen": -0.006989480461925268,
|
| 1395 |
+
"rewards/margins": 0.010014806874096394,
|
| 1396 |
+
"rewards/rejected": -0.017004288733005524,
|
| 1397 |
+
"step": 890
|
| 1398 |
+
},
|
| 1399 |
+
{
|
| 1400 |
+
"epoch": 1.88,
|
| 1401 |
+
"learning_rate": 4.870879364444109e-08,
|
| 1402 |
+
"logits/chosen": 0.8013142347335815,
|
| 1403 |
+
"logits/rejected": 0.8696510195732117,
|
| 1404 |
+
"logps/chosen": -252.10507202148438,
|
| 1405 |
+
"logps/rejected": -206.7936553955078,
|
| 1406 |
+
"loss": 0.0021,
|
| 1407 |
+
"rewards/accuracies": 0.6499999761581421,
|
| 1408 |
+
"rewards/chosen": -0.007737092673778534,
|
| 1409 |
+
"rewards/margins": 0.01070366334170103,
|
| 1410 |
+
"rewards/rejected": -0.01844075694680214,
|
| 1411 |
+
"step": 900
|
| 1412 |
+
},
|
| 1413 |
+
{
|
| 1414 |
+
"epoch": 1.88,
|
| 1415 |
+
"eval_logits/chosen": 0.7806341648101807,
|
| 1416 |
+
"eval_logits/rejected": 0.8832870721817017,
|
| 1417 |
+
"eval_logps/chosen": -261.0921936035156,
|
| 1418 |
+
"eval_logps/rejected": -233.21180725097656,
|
| 1419 |
+
"eval_loss": 0.002123194048181176,
|
| 1420 |
+
"eval_rewards/accuracies": 0.6899999976158142,
|
| 1421 |
+
"eval_rewards/chosen": -0.008480267599225044,
|
| 1422 |
+
"eval_rewards/margins": 0.009276580065488815,
|
| 1423 |
+
"eval_rewards/rejected": -0.01775684952735901,
|
| 1424 |
+
"eval_runtime": 325.0149,
|
| 1425 |
+
"eval_samples_per_second": 6.154,
|
| 1426 |
+
"eval_steps_per_second": 0.385,
|
| 1427 |
+
"step": 900
|
| 1428 |
+
},
|
| 1429 |
+
{
|
| 1430 |
+
"epoch": 1.9,
|
| 1431 |
+
"learning_rate": 3.237434340521789e-08,
|
| 1432 |
+
"logits/chosen": 0.7978643774986267,
|
| 1433 |
+
"logits/rejected": 0.8687127828598022,
|
| 1434 |
+
"logps/chosen": -263.38275146484375,
|
| 1435 |
+
"logps/rejected": -247.8026123046875,
|
| 1436 |
+
"loss": 0.0021,
|
| 1437 |
+
"rewards/accuracies": 0.668749988079071,
|
| 1438 |
+
"rewards/chosen": -0.007231117691844702,
|
| 1439 |
+
"rewards/margins": 0.009652243927121162,
|
| 1440 |
+
"rewards/rejected": -0.016883360221982002,
|
| 1441 |
+
"step": 910
|
| 1442 |
+
},
|
| 1443 |
+
{
|
| 1444 |
+
"epoch": 1.93,
|
| 1445 |
+
"learning_rate": 1.93478202307823e-08,
|
| 1446 |
+
"logits/chosen": 0.7963850498199463,
|
| 1447 |
+
"logits/rejected": 0.8160678148269653,
|
| 1448 |
+
"logps/chosen": -242.1365966796875,
|
| 1449 |
+
"logps/rejected": -246.0305938720703,
|
| 1450 |
+
"loss": 0.0021,
|
| 1451 |
+
"rewards/accuracies": 0.643750011920929,
|
| 1452 |
+
"rewards/chosen": -0.0077395932748913765,
|
| 1453 |
+
"rewards/margins": 0.007968437857925892,
|
| 1454 |
+
"rewards/rejected": -0.01570803113281727,
|
| 1455 |
+
"step": 920
|
| 1456 |
+
},
|
| 1457 |
+
{
|
| 1458 |
+
"epoch": 1.95,
|
| 1459 |
+
"learning_rate": 9.646686570697062e-09,
|
| 1460 |
+
"logits/chosen": 0.862303614616394,
|
| 1461 |
+
"logits/rejected": 0.8678015470504761,
|
| 1462 |
+
"logps/chosen": -257.33099365234375,
|
| 1463 |
+
"logps/rejected": -249.9061737060547,
|
| 1464 |
+
"loss": 0.0021,
|
| 1465 |
+
"rewards/accuracies": 0.7124999761581421,
|
| 1466 |
+
"rewards/chosen": -0.007601047400385141,
|
| 1467 |
+
"rewards/margins": 0.009036187082529068,
|
| 1468 |
+
"rewards/rejected": -0.01663723587989807,
|
| 1469 |
+
"step": 930
|
| 1470 |
+
},
|
| 1471 |
+
{
|
| 1472 |
+
"epoch": 1.97,
|
| 1473 |
+
"learning_rate": 3.283947088983663e-09,
|
| 1474 |
+
"logits/chosen": 0.8371657133102417,
|
| 1475 |
+
"logits/rejected": 0.8322643041610718,
|
| 1476 |
+
"logps/chosen": -238.14657592773438,
|
| 1477 |
+
"logps/rejected": -243.3525390625,
|
| 1478 |
+
"loss": 0.0021,
|
| 1479 |
+
"rewards/accuracies": 0.65625,
|
| 1480 |
+
"rewards/chosen": -0.009062298573553562,
|
| 1481 |
+
"rewards/margins": 0.008633644320070744,
|
| 1482 |
+
"rewards/rejected": -0.017695942893624306,
|
| 1483 |
+
"step": 940
|
| 1484 |
+
},
|
| 1485 |
+
{
|
| 1486 |
+
"epoch": 1.99,
|
| 1487 |
+
"learning_rate": 2.681312309735229e-10,
|
| 1488 |
+
"logits/chosen": 0.8020931482315063,
|
| 1489 |
+
"logits/rejected": 0.9026565551757812,
|
| 1490 |
+
"logps/chosen": -231.6744842529297,
|
| 1491 |
+
"logps/rejected": -229.53726196289062,
|
| 1492 |
+
"loss": 0.0021,
|
| 1493 |
+
"rewards/accuracies": 0.612500011920929,
|
| 1494 |
+
"rewards/chosen": -0.007398143410682678,
|
| 1495 |
+
"rewards/margins": 0.009477959014475346,
|
| 1496 |
+
"rewards/rejected": -0.0168761033564806,
|
| 1497 |
+
"step": 950
|
| 1498 |
+
},
|
| 1499 |
+
{
|
| 1500 |
+
"epoch": 2.0,
|
| 1501 |
+
"step": 954,
|
| 1502 |
+
"total_flos": 0.0,
|
| 1503 |
+
"train_loss": 0.0021909422920118682,
|
| 1504 |
+
"train_runtime": 18127.9992,
|
| 1505 |
+
"train_samples_per_second": 3.372,
|
| 1506 |
+
"train_steps_per_second": 0.053
|
| 1507 |
+
}
|
| 1508 |
+
],
|
| 1509 |
+
"logging_steps": 10,
|
| 1510 |
+
"max_steps": 954,
|
| 1511 |
+
"num_input_tokens_seen": 0,
|
| 1512 |
+
"num_train_epochs": 2,
|
| 1513 |
+
"save_steps": 100,
|
| 1514 |
+
"total_flos": 0.0,
|
| 1515 |
+
"train_batch_size": 4,
|
| 1516 |
+
"trial_name": null,
|
| 1517 |
+
"trial_params": null
|
| 1518 |
+
}
|