Model save
Browse files- README.md +77 -0
- adapter_model.safetensors +1 -1
- all_results.json +21 -0
- eval_results.json +16 -0
- runs/Jan18_15-28-29_08134be46a59/events.out.tfevents.1705591929.08134be46a59.1750.0 +2 -2
- runs/Jan18_15-28-29_08134be46a59/events.out.tfevents.1705609138.08134be46a59.1750.1 +3 -0
- train_results.json +8 -0
- trainer_state.json +1538 -0
README.md
ADDED
|
@@ -0,0 +1,77 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
license: apache-2.0
|
| 3 |
+
library_name: peft
|
| 4 |
+
tags:
|
| 5 |
+
- trl
|
| 6 |
+
- dpo
|
| 7 |
+
- generated_from_trainer
|
| 8 |
+
base_model: sablo/sablo-pebble-mistral
|
| 9 |
+
model-index:
|
| 10 |
+
- name: sablo-pebble-mistral-dpo-lora-HelpSteer_binarized
|
| 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 |
+
# sablo-pebble-mistral-dpo-lora-HelpSteer_binarized
|
| 18 |
+
|
| 19 |
+
This model is a fine-tuned version of [sablo/sablo-pebble-mistral](https://huggingface.co/sablo/sablo-pebble-mistral) on the None dataset.
|
| 20 |
+
It achieves the following results on the evaluation set:
|
| 21 |
+
- Loss: 0.5212
|
| 22 |
+
- Rewards/chosen: -2.5398
|
| 23 |
+
- Rewards/rejected: -3.5311
|
| 24 |
+
- Rewards/accuracies: 0.7406
|
| 25 |
+
- Rewards/margins: 0.9913
|
| 26 |
+
- Logps/rejected: -214.9349
|
| 27 |
+
- Logps/chosen: -206.5847
|
| 28 |
+
- Logits/rejected: -2.0624
|
| 29 |
+
- Logits/chosen: -2.1620
|
| 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: 8
|
| 51 |
+
- seed: 42
|
| 52 |
+
- distributed_type: multi-GPU
|
| 53 |
+
- gradient_accumulation_steps: 2
|
| 54 |
+
- total_train_batch_size: 8
|
| 55 |
+
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
|
| 56 |
+
- lr_scheduler_type: cosine
|
| 57 |
+
- lr_scheduler_warmup_ratio: 0.1
|
| 58 |
+
- num_epochs: 1
|
| 59 |
+
|
| 60 |
+
### Training results
|
| 61 |
+
|
| 62 |
+
| Training Loss | Epoch | Step | Validation Loss | Rewards/chosen | Rewards/rejected | Rewards/accuracies | Rewards/margins | Logps/rejected | Logps/chosen | Logits/rejected | Logits/chosen |
|
| 63 |
+
|:-------------:|:-----:|:----:|:---------------:|:--------------:|:----------------:|:------------------:|:---------------:|:--------------:|:------------:|:---------------:|:-------------:|
|
| 64 |
+
| 0.652 | 0.2 | 200 | 0.6595 | 0.0498 | -0.0320 | 0.6415 | 0.0818 | -98.2975 | -120.2629 | -2.0236 | -2.1231 |
|
| 65 |
+
| 0.4905 | 0.39 | 400 | 0.5551 | -1.6581 | -2.1527 | 0.6958 | 0.4946 | -168.9884 | -177.1946 | -2.0950 | -2.1951 |
|
| 66 |
+
| 0.4249 | 0.59 | 600 | 0.5327 | -3.4554 | -4.3247 | 0.7241 | 0.8693 | -241.3867 | -237.1045 | -2.0782 | -2.1773 |
|
| 67 |
+
| 0.5858 | 0.79 | 800 | 0.5207 | -2.5072 | -3.4512 | 0.7335 | 0.9440 | -212.2718 | -205.4982 | -2.0586 | -2.1591 |
|
| 68 |
+
| 0.6128 | 0.98 | 1000 | 0.5212 | -2.5398 | -3.5311 | 0.7406 | 0.9913 | -214.9349 | -206.5847 | -2.0624 | -2.1620 |
|
| 69 |
+
|
| 70 |
+
|
| 71 |
+
### Framework versions
|
| 72 |
+
|
| 73 |
+
- PEFT 0.7.1
|
| 74 |
+
- Transformers 4.36.2
|
| 75 |
+
- Pytorch 2.0.1+cu118
|
| 76 |
+
- Datasets 2.14.6
|
| 77 |
+
- Tokenizers 0.15.0
|
adapter_model.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 83945744
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:5595d5709e08dfe18a5e6f2986317e21a3e06942bd5a4d3e5c0bbc18374bcd7e
|
| 3 |
size 83945744
|
all_results.json
ADDED
|
@@ -0,0 +1,21 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 1.0,
|
| 3 |
+
"eval_logits/chosen": -2.1620335578918457,
|
| 4 |
+
"eval_logits/rejected": -2.062356948852539,
|
| 5 |
+
"eval_logps/chosen": -206.584716796875,
|
| 6 |
+
"eval_logps/rejected": -214.93492126464844,
|
| 7 |
+
"eval_loss": 0.521207869052887,
|
| 8 |
+
"eval_rewards/accuracies": 0.7405660152435303,
|
| 9 |
+
"eval_rewards/chosen": -2.5398268699645996,
|
| 10 |
+
"eval_rewards/margins": 0.9912916421890259,
|
| 11 |
+
"eval_rewards/rejected": -3.531118631362915,
|
| 12 |
+
"eval_runtime": 424.2832,
|
| 13 |
+
"eval_samples": 418,
|
| 14 |
+
"eval_samples_per_second": 0.985,
|
| 15 |
+
"eval_steps_per_second": 0.125,
|
| 16 |
+
"train_loss": 0.5583291621658746,
|
| 17 |
+
"train_runtime": 16785.8838,
|
| 18 |
+
"train_samples": 8130,
|
| 19 |
+
"train_samples_per_second": 0.484,
|
| 20 |
+
"train_steps_per_second": 0.061
|
| 21 |
+
}
|
eval_results.json
ADDED
|
@@ -0,0 +1,16 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 1.0,
|
| 3 |
+
"eval_logits/chosen": -2.1620335578918457,
|
| 4 |
+
"eval_logits/rejected": -2.062356948852539,
|
| 5 |
+
"eval_logps/chosen": -206.584716796875,
|
| 6 |
+
"eval_logps/rejected": -214.93492126464844,
|
| 7 |
+
"eval_loss": 0.521207869052887,
|
| 8 |
+
"eval_rewards/accuracies": 0.7405660152435303,
|
| 9 |
+
"eval_rewards/chosen": -2.5398268699645996,
|
| 10 |
+
"eval_rewards/margins": 0.9912916421890259,
|
| 11 |
+
"eval_rewards/rejected": -3.531118631362915,
|
| 12 |
+
"eval_runtime": 424.2832,
|
| 13 |
+
"eval_samples": 418,
|
| 14 |
+
"eval_samples_per_second": 0.985,
|
| 15 |
+
"eval_steps_per_second": 0.125
|
| 16 |
+
}
|
runs/Jan18_15-28-29_08134be46a59/events.out.tfevents.1705591929.08134be46a59.1750.0
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:a48003c4bae929ea8ea233e41f4ee0d6502473290e7d839752a5e0f4387e7e89
|
| 3 |
+
size 73255
|
runs/Jan18_15-28-29_08134be46a59/events.out.tfevents.1705609138.08134be46a59.1750.1
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:00be72d7c7f7977576a9ef9c79387fdd23171778aeb07fa3ceb8cd5e2946f294
|
| 3 |
+
size 828
|
train_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 1.0,
|
| 3 |
+
"train_loss": 0.5583291621658746,
|
| 4 |
+
"train_runtime": 16785.8838,
|
| 5 |
+
"train_samples": 8130,
|
| 6 |
+
"train_samples_per_second": 0.484,
|
| 7 |
+
"train_steps_per_second": 0.061
|
| 8 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,1538 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 0.999508116084604,
|
| 5 |
+
"eval_steps": 200,
|
| 6 |
+
"global_step": 1016,
|
| 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": 4.901960784313726e-08,
|
| 14 |
+
"logits/chosen": -2.0737838745117188,
|
| 15 |
+
"logits/rejected": -2.1456010341644287,
|
| 16 |
+
"logps/chosen": -95.6572265625,
|
| 17 |
+
"logps/rejected": -106.55765533447266,
|
| 18 |
+
"loss": 0.6931,
|
| 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": 0.01,
|
| 27 |
+
"learning_rate": 4.901960784313725e-07,
|
| 28 |
+
"logits/chosen": -2.165830373764038,
|
| 29 |
+
"logits/rejected": -2.060776948928833,
|
| 30 |
+
"logps/chosen": -121.03773498535156,
|
| 31 |
+
"logps/rejected": -87.5294189453125,
|
| 32 |
+
"loss": 0.6918,
|
| 33 |
+
"rewards/accuracies": 0.5,
|
| 34 |
+
"rewards/chosen": 0.00047249632189050317,
|
| 35 |
+
"rewards/margins": 0.002704059472307563,
|
| 36 |
+
"rewards/rejected": -0.002231562975794077,
|
| 37 |
+
"step": 10
|
| 38 |
+
},
|
| 39 |
+
{
|
| 40 |
+
"epoch": 0.02,
|
| 41 |
+
"learning_rate": 9.80392156862745e-07,
|
| 42 |
+
"logits/chosen": -2.1349050998687744,
|
| 43 |
+
"logits/rejected": -2.016066312789917,
|
| 44 |
+
"logps/chosen": -130.94175720214844,
|
| 45 |
+
"logps/rejected": -105.7674789428711,
|
| 46 |
+
"loss": 0.6935,
|
| 47 |
+
"rewards/accuracies": 0.48750001192092896,
|
| 48 |
+
"rewards/chosen": -0.0016262540593743324,
|
| 49 |
+
"rewards/margins": -0.0006325626163743436,
|
| 50 |
+
"rewards/rejected": -0.0009936915012076497,
|
| 51 |
+
"step": 20
|
| 52 |
+
},
|
| 53 |
+
{
|
| 54 |
+
"epoch": 0.03,
|
| 55 |
+
"learning_rate": 1.4705882352941177e-06,
|
| 56 |
+
"logits/chosen": -2.2818872928619385,
|
| 57 |
+
"logits/rejected": -2.1805636882781982,
|
| 58 |
+
"logps/chosen": -121.03263854980469,
|
| 59 |
+
"logps/rejected": -104.84712982177734,
|
| 60 |
+
"loss": 0.6942,
|
| 61 |
+
"rewards/accuracies": 0.3499999940395355,
|
| 62 |
+
"rewards/chosen": -0.003329185303300619,
|
| 63 |
+
"rewards/margins": -0.002142944373190403,
|
| 64 |
+
"rewards/rejected": -0.0011862408136948943,
|
| 65 |
+
"step": 30
|
| 66 |
+
},
|
| 67 |
+
{
|
| 68 |
+
"epoch": 0.04,
|
| 69 |
+
"learning_rate": 1.96078431372549e-06,
|
| 70 |
+
"logits/chosen": -2.2757694721221924,
|
| 71 |
+
"logits/rejected": -2.156691074371338,
|
| 72 |
+
"logps/chosen": -126.5389633178711,
|
| 73 |
+
"logps/rejected": -105.20024108886719,
|
| 74 |
+
"loss": 0.6919,
|
| 75 |
+
"rewards/accuracies": 0.6000000238418579,
|
| 76 |
+
"rewards/chosen": 0.0010745985200628638,
|
| 77 |
+
"rewards/margins": 0.002502765040844679,
|
| 78 |
+
"rewards/rejected": -0.0014281660551205277,
|
| 79 |
+
"step": 40
|
| 80 |
+
},
|
| 81 |
+
{
|
| 82 |
+
"epoch": 0.05,
|
| 83 |
+
"learning_rate": 2.450980392156863e-06,
|
| 84 |
+
"logits/chosen": -2.3002381324768066,
|
| 85 |
+
"logits/rejected": -2.206784725189209,
|
| 86 |
+
"logps/chosen": -124.18415832519531,
|
| 87 |
+
"logps/rejected": -98.63652801513672,
|
| 88 |
+
"loss": 0.6916,
|
| 89 |
+
"rewards/accuracies": 0.612500011920929,
|
| 90 |
+
"rewards/chosen": 0.0012203993974253535,
|
| 91 |
+
"rewards/margins": 0.003048995044082403,
|
| 92 |
+
"rewards/rejected": -0.0018285956466570497,
|
| 93 |
+
"step": 50
|
| 94 |
+
},
|
| 95 |
+
{
|
| 96 |
+
"epoch": 0.06,
|
| 97 |
+
"learning_rate": 2.9411764705882355e-06,
|
| 98 |
+
"logits/chosen": -2.3112730979919434,
|
| 99 |
+
"logits/rejected": -2.232532501220703,
|
| 100 |
+
"logps/chosen": -126.93055725097656,
|
| 101 |
+
"logps/rejected": -109.7610092163086,
|
| 102 |
+
"loss": 0.6914,
|
| 103 |
+
"rewards/accuracies": 0.6000000238418579,
|
| 104 |
+
"rewards/chosen": 0.01018393412232399,
|
| 105 |
+
"rewards/margins": 0.0034624538384377956,
|
| 106 |
+
"rewards/rejected": 0.006721480283886194,
|
| 107 |
+
"step": 60
|
| 108 |
+
},
|
| 109 |
+
{
|
| 110 |
+
"epoch": 0.07,
|
| 111 |
+
"learning_rate": 3.431372549019608e-06,
|
| 112 |
+
"logits/chosen": -2.2930376529693604,
|
| 113 |
+
"logits/rejected": -2.187328815460205,
|
| 114 |
+
"logps/chosen": -119.68111419677734,
|
| 115 |
+
"logps/rejected": -93.93470001220703,
|
| 116 |
+
"loss": 0.6912,
|
| 117 |
+
"rewards/accuracies": 0.5874999761581421,
|
| 118 |
+
"rewards/chosen": 0.007645626552402973,
|
| 119 |
+
"rewards/margins": 0.003846182022243738,
|
| 120 |
+
"rewards/rejected": 0.0037994447629898787,
|
| 121 |
+
"step": 70
|
| 122 |
+
},
|
| 123 |
+
{
|
| 124 |
+
"epoch": 0.08,
|
| 125 |
+
"learning_rate": 3.92156862745098e-06,
|
| 126 |
+
"logits/chosen": -2.144854784011841,
|
| 127 |
+
"logits/rejected": -2.016728401184082,
|
| 128 |
+
"logps/chosen": -132.2415313720703,
|
| 129 |
+
"logps/rejected": -106.7207260131836,
|
| 130 |
+
"loss": 0.6902,
|
| 131 |
+
"rewards/accuracies": 0.637499988079071,
|
| 132 |
+
"rewards/chosen": 0.015773242339491844,
|
| 133 |
+
"rewards/margins": 0.006012483034282923,
|
| 134 |
+
"rewards/rejected": 0.009760759770870209,
|
| 135 |
+
"step": 80
|
| 136 |
+
},
|
| 137 |
+
{
|
| 138 |
+
"epoch": 0.09,
|
| 139 |
+
"learning_rate": 4.411764705882353e-06,
|
| 140 |
+
"logits/chosen": -2.095534324645996,
|
| 141 |
+
"logits/rejected": -1.9636192321777344,
|
| 142 |
+
"logps/chosen": -106.81976318359375,
|
| 143 |
+
"logps/rejected": -83.68408966064453,
|
| 144 |
+
"loss": 0.6887,
|
| 145 |
+
"rewards/accuracies": 0.6000000238418579,
|
| 146 |
+
"rewards/chosen": 0.02147563174366951,
|
| 147 |
+
"rewards/margins": 0.009103062562644482,
|
| 148 |
+
"rewards/rejected": 0.012372570112347603,
|
| 149 |
+
"step": 90
|
| 150 |
+
},
|
| 151 |
+
{
|
| 152 |
+
"epoch": 0.1,
|
| 153 |
+
"learning_rate": 4.901960784313726e-06,
|
| 154 |
+
"logits/chosen": -2.0606656074523926,
|
| 155 |
+
"logits/rejected": -1.87876296043396,
|
| 156 |
+
"logps/chosen": -141.498779296875,
|
| 157 |
+
"logps/rejected": -105.37713623046875,
|
| 158 |
+
"loss": 0.6811,
|
| 159 |
+
"rewards/accuracies": 0.75,
|
| 160 |
+
"rewards/chosen": 0.04738330841064453,
|
| 161 |
+
"rewards/margins": 0.024673232808709145,
|
| 162 |
+
"rewards/rejected": 0.022710075601935387,
|
| 163 |
+
"step": 100
|
| 164 |
+
},
|
| 165 |
+
{
|
| 166 |
+
"epoch": 0.11,
|
| 167 |
+
"learning_rate": 4.9990549169459415e-06,
|
| 168 |
+
"logits/chosen": -2.2754921913146973,
|
| 169 |
+
"logits/rejected": -2.135282516479492,
|
| 170 |
+
"logps/chosen": -124.19816589355469,
|
| 171 |
+
"logps/rejected": -98.95077514648438,
|
| 172 |
+
"loss": 0.6777,
|
| 173 |
+
"rewards/accuracies": 0.7250000238418579,
|
| 174 |
+
"rewards/chosen": 0.056061066687107086,
|
| 175 |
+
"rewards/margins": 0.03164363652467728,
|
| 176 |
+
"rewards/rejected": 0.02441743016242981,
|
| 177 |
+
"step": 110
|
| 178 |
+
},
|
| 179 |
+
{
|
| 180 |
+
"epoch": 0.12,
|
| 181 |
+
"learning_rate": 4.995216741642263e-06,
|
| 182 |
+
"logits/chosen": -2.280646324157715,
|
| 183 |
+
"logits/rejected": -2.206347942352295,
|
| 184 |
+
"logps/chosen": -115.8212890625,
|
| 185 |
+
"logps/rejected": -96.82814025878906,
|
| 186 |
+
"loss": 0.6805,
|
| 187 |
+
"rewards/accuracies": 0.6625000238418579,
|
| 188 |
+
"rewards/chosen": 0.07867949455976486,
|
| 189 |
+
"rewards/margins": 0.02664627507328987,
|
| 190 |
+
"rewards/rejected": 0.05203322693705559,
|
| 191 |
+
"step": 120
|
| 192 |
+
},
|
| 193 |
+
{
|
| 194 |
+
"epoch": 0.13,
|
| 195 |
+
"learning_rate": 4.988430936991089e-06,
|
| 196 |
+
"logits/chosen": -2.2835030555725098,
|
| 197 |
+
"logits/rejected": -2.140094041824341,
|
| 198 |
+
"logps/chosen": -127.94944763183594,
|
| 199 |
+
"logps/rejected": -101.47581481933594,
|
| 200 |
+
"loss": 0.6635,
|
| 201 |
+
"rewards/accuracies": 0.75,
|
| 202 |
+
"rewards/chosen": 0.10174749791622162,
|
| 203 |
+
"rewards/margins": 0.06262228637933731,
|
| 204 |
+
"rewards/rejected": 0.039125215262174606,
|
| 205 |
+
"step": 130
|
| 206 |
+
},
|
| 207 |
+
{
|
| 208 |
+
"epoch": 0.14,
|
| 209 |
+
"learning_rate": 4.978705519144525e-06,
|
| 210 |
+
"logits/chosen": -2.18971586227417,
|
| 211 |
+
"logits/rejected": -2.0279135704040527,
|
| 212 |
+
"logps/chosen": -140.58509826660156,
|
| 213 |
+
"logps/rejected": -104.29924011230469,
|
| 214 |
+
"loss": 0.664,
|
| 215 |
+
"rewards/accuracies": 0.75,
|
| 216 |
+
"rewards/chosen": 0.07688155025243759,
|
| 217 |
+
"rewards/margins": 0.061921559274196625,
|
| 218 |
+
"rewards/rejected": 0.014959996566176414,
|
| 219 |
+
"step": 140
|
| 220 |
+
},
|
| 221 |
+
{
|
| 222 |
+
"epoch": 0.15,
|
| 223 |
+
"learning_rate": 4.966051976854862e-06,
|
| 224 |
+
"logits/chosen": -2.333808183670044,
|
| 225 |
+
"logits/rejected": -2.2144458293914795,
|
| 226 |
+
"logps/chosen": -111.2592544555664,
|
| 227 |
+
"logps/rejected": -89.93299865722656,
|
| 228 |
+
"loss": 0.6526,
|
| 229 |
+
"rewards/accuracies": 0.762499988079071,
|
| 230 |
+
"rewards/chosen": 0.11892716586589813,
|
| 231 |
+
"rewards/margins": 0.08827908337116241,
|
| 232 |
+
"rewards/rejected": 0.030648082494735718,
|
| 233 |
+
"step": 150
|
| 234 |
+
},
|
| 235 |
+
{
|
| 236 |
+
"epoch": 0.16,
|
| 237 |
+
"learning_rate": 4.950485257902782e-06,
|
| 238 |
+
"logits/chosen": -2.209681749343872,
|
| 239 |
+
"logits/rejected": -2.1354687213897705,
|
| 240 |
+
"logps/chosen": -122.34986877441406,
|
| 241 |
+
"logps/rejected": -97.61170959472656,
|
| 242 |
+
"loss": 0.6622,
|
| 243 |
+
"rewards/accuracies": 0.637499988079071,
|
| 244 |
+
"rewards/chosen": 0.09915992617607117,
|
| 245 |
+
"rewards/margins": 0.06800667941570282,
|
| 246 |
+
"rewards/rejected": 0.031153246760368347,
|
| 247 |
+
"step": 160
|
| 248 |
+
},
|
| 249 |
+
{
|
| 250 |
+
"epoch": 0.17,
|
| 251 |
+
"learning_rate": 4.932023751439358e-06,
|
| 252 |
+
"logits/chosen": -2.276695489883423,
|
| 253 |
+
"logits/rejected": -2.118220329284668,
|
| 254 |
+
"logps/chosen": -131.77114868164062,
|
| 255 |
+
"logps/rejected": -103.4157943725586,
|
| 256 |
+
"loss": 0.6666,
|
| 257 |
+
"rewards/accuracies": 0.6499999761581421,
|
| 258 |
+
"rewards/chosen": 0.10167907178401947,
|
| 259 |
+
"rewards/margins": 0.06214705854654312,
|
| 260 |
+
"rewards/rejected": 0.039532024413347244,
|
| 261 |
+
"step": 170
|
| 262 |
+
},
|
| 263 |
+
{
|
| 264 |
+
"epoch": 0.18,
|
| 265 |
+
"learning_rate": 4.9106892662627395e-06,
|
| 266 |
+
"logits/chosen": -2.347627878189087,
|
| 267 |
+
"logits/rejected": -2.223806858062744,
|
| 268 |
+
"logps/chosen": -125.82316589355469,
|
| 269 |
+
"logps/rejected": -102.32342529296875,
|
| 270 |
+
"loss": 0.6605,
|
| 271 |
+
"rewards/accuracies": 0.7124999761581421,
|
| 272 |
+
"rewards/chosen": 0.13379618525505066,
|
| 273 |
+
"rewards/margins": 0.07378261536359787,
|
| 274 |
+
"rewards/rejected": 0.06001356244087219,
|
| 275 |
+
"step": 180
|
| 276 |
+
},
|
| 277 |
+
{
|
| 278 |
+
"epoch": 0.19,
|
| 279 |
+
"learning_rate": 4.886507005055149e-06,
|
| 280 |
+
"logits/chosen": -2.299999713897705,
|
| 281 |
+
"logits/rejected": -2.124567985534668,
|
| 282 |
+
"logps/chosen": -135.9125518798828,
|
| 283 |
+
"logps/rejected": -102.34548950195312,
|
| 284 |
+
"loss": 0.6471,
|
| 285 |
+
"rewards/accuracies": 0.762499988079071,
|
| 286 |
+
"rewards/chosen": 0.12664374709129333,
|
| 287 |
+
"rewards/margins": 0.10224989801645279,
|
| 288 |
+
"rewards/rejected": 0.024393849074840546,
|
| 289 |
+
"step": 190
|
| 290 |
+
},
|
| 291 |
+
{
|
| 292 |
+
"epoch": 0.2,
|
| 293 |
+
"learning_rate": 4.859505534610658e-06,
|
| 294 |
+
"logits/chosen": -2.2595176696777344,
|
| 295 |
+
"logits/rejected": -2.167226552963257,
|
| 296 |
+
"logps/chosen": -115.63578033447266,
|
| 297 |
+
"logps/rejected": -97.40113830566406,
|
| 298 |
+
"loss": 0.652,
|
| 299 |
+
"rewards/accuracies": 0.6625000238418579,
|
| 300 |
+
"rewards/chosen": 0.07887722551822662,
|
| 301 |
+
"rewards/margins": 0.09930779039859772,
|
| 302 |
+
"rewards/rejected": -0.020430563017725945,
|
| 303 |
+
"step": 200
|
| 304 |
+
},
|
| 305 |
+
{
|
| 306 |
+
"epoch": 0.2,
|
| 307 |
+
"eval_logits/chosen": -2.1231298446655273,
|
| 308 |
+
"eval_logits/rejected": -2.0235936641693115,
|
| 309 |
+
"eval_logps/chosen": -120.2629165649414,
|
| 310 |
+
"eval_logps/rejected": -98.29754638671875,
|
| 311 |
+
"eval_loss": 0.6594940423965454,
|
| 312 |
+
"eval_rewards/accuracies": 0.6415094137191772,
|
| 313 |
+
"eval_rewards/chosen": 0.04982735216617584,
|
| 314 |
+
"eval_rewards/margins": 0.08182442933320999,
|
| 315 |
+
"eval_rewards/rejected": -0.031997084617614746,
|
| 316 |
+
"eval_runtime": 417.8564,
|
| 317 |
+
"eval_samples_per_second": 1.0,
|
| 318 |
+
"eval_steps_per_second": 0.127,
|
| 319 |
+
"step": 200
|
| 320 |
+
},
|
| 321 |
+
{
|
| 322 |
+
"epoch": 0.21,
|
| 323 |
+
"learning_rate": 4.829716752088893e-06,
|
| 324 |
+
"logits/chosen": -2.202777147293091,
|
| 325 |
+
"logits/rejected": -2.1421408653259277,
|
| 326 |
+
"logps/chosen": -107.44587707519531,
|
| 327 |
+
"logps/rejected": -101.29158020019531,
|
| 328 |
+
"loss": 0.6594,
|
| 329 |
+
"rewards/accuracies": 0.6875,
|
| 330 |
+
"rewards/chosen": 0.03508143126964569,
|
| 331 |
+
"rewards/margins": 0.07955195009708405,
|
| 332 |
+
"rewards/rejected": -0.04447052255272865,
|
| 333 |
+
"step": 210
|
| 334 |
+
},
|
| 335 |
+
{
|
| 336 |
+
"epoch": 0.22,
|
| 337 |
+
"learning_rate": 4.797175847334535e-06,
|
| 338 |
+
"logits/chosen": -2.217074155807495,
|
| 339 |
+
"logits/rejected": -2.0960960388183594,
|
| 340 |
+
"logps/chosen": -130.84152221679688,
|
| 341 |
+
"logps/rejected": -107.61418151855469,
|
| 342 |
+
"loss": 0.6708,
|
| 343 |
+
"rewards/accuracies": 0.6499999761581421,
|
| 344 |
+
"rewards/chosen": 0.0030643828213214874,
|
| 345 |
+
"rewards/margins": 0.05633828788995743,
|
| 346 |
+
"rewards/rejected": -0.05327390506863594,
|
| 347 |
+
"step": 220
|
| 348 |
+
},
|
| 349 |
+
{
|
| 350 |
+
"epoch": 0.23,
|
| 351 |
+
"learning_rate": 4.761921261307143e-06,
|
| 352 |
+
"logits/chosen": -2.2311947345733643,
|
| 353 |
+
"logits/rejected": -2.0887582302093506,
|
| 354 |
+
"logps/chosen": -123.64229583740234,
|
| 355 |
+
"logps/rejected": -103.31034851074219,
|
| 356 |
+
"loss": 0.655,
|
| 357 |
+
"rewards/accuracies": 0.6875,
|
| 358 |
+
"rewards/chosen": -0.009160916320979595,
|
| 359 |
+
"rewards/margins": 0.09632667899131775,
|
| 360 |
+
"rewards/rejected": -0.10548758506774902,
|
| 361 |
+
"step": 230
|
| 362 |
+
},
|
| 363 |
+
{
|
| 364 |
+
"epoch": 0.24,
|
| 365 |
+
"learning_rate": 4.723994640670377e-06,
|
| 366 |
+
"logits/chosen": -2.2253684997558594,
|
| 367 |
+
"logits/rejected": -2.0454909801483154,
|
| 368 |
+
"logps/chosen": -137.86196899414062,
|
| 369 |
+
"logps/rejected": -104.11041259765625,
|
| 370 |
+
"loss": 0.6146,
|
| 371 |
+
"rewards/accuracies": 0.8125,
|
| 372 |
+
"rewards/chosen": -0.011105505749583244,
|
| 373 |
+
"rewards/margins": 0.1803070306777954,
|
| 374 |
+
"rewards/rejected": -0.1914125233888626,
|
| 375 |
+
"step": 240
|
| 376 |
+
},
|
| 377 |
+
{
|
| 378 |
+
"epoch": 0.25,
|
| 379 |
+
"learning_rate": 4.68344078859431e-06,
|
| 380 |
+
"logits/chosen": -2.1830849647521973,
|
| 381 |
+
"logits/rejected": -2.1306838989257812,
|
| 382 |
+
"logps/chosen": -119.04063415527344,
|
| 383 |
+
"logps/rejected": -113.45658111572266,
|
| 384 |
+
"loss": 0.6407,
|
| 385 |
+
"rewards/accuracies": 0.6499999761581421,
|
| 386 |
+
"rewards/chosen": -0.10984460264444351,
|
| 387 |
+
"rewards/margins": 0.1269720494747162,
|
| 388 |
+
"rewards/rejected": -0.2368166148662567,
|
| 389 |
+
"step": 250
|
| 390 |
+
},
|
| 391 |
+
{
|
| 392 |
+
"epoch": 0.26,
|
| 393 |
+
"learning_rate": 4.6403076118289006e-06,
|
| 394 |
+
"logits/chosen": -2.151690721511841,
|
| 395 |
+
"logits/rejected": -2.0012199878692627,
|
| 396 |
+
"logps/chosen": -135.11477661132812,
|
| 397 |
+
"logps/rejected": -110.16157531738281,
|
| 398 |
+
"loss": 0.6217,
|
| 399 |
+
"rewards/accuracies": 0.699999988079071,
|
| 400 |
+
"rewards/chosen": -0.162130668759346,
|
| 401 |
+
"rewards/margins": 0.17287474870681763,
|
| 402 |
+
"rewards/rejected": -0.33500543236732483,
|
| 403 |
+
"step": 260
|
| 404 |
+
},
|
| 405 |
+
{
|
| 406 |
+
"epoch": 0.27,
|
| 407 |
+
"learning_rate": 4.5946460641111776e-06,
|
| 408 |
+
"logits/chosen": -2.3390426635742188,
|
| 409 |
+
"logits/rejected": -2.1155288219451904,
|
| 410 |
+
"logps/chosen": -132.8888397216797,
|
| 411 |
+
"logps/rejected": -110.8713607788086,
|
| 412 |
+
"loss": 0.5927,
|
| 413 |
+
"rewards/accuracies": 0.75,
|
| 414 |
+
"rewards/chosen": -0.19292452931404114,
|
| 415 |
+
"rewards/margins": 0.24748054146766663,
|
| 416 |
+
"rewards/rejected": -0.44040507078170776,
|
| 417 |
+
"step": 270
|
| 418 |
+
},
|
| 419 |
+
{
|
| 420 |
+
"epoch": 0.28,
|
| 421 |
+
"learning_rate": 4.546510085972983e-06,
|
| 422 |
+
"logits/chosen": -2.3495311737060547,
|
| 423 |
+
"logits/rejected": -2.2263104915618896,
|
| 424 |
+
"logps/chosen": -143.91481018066406,
|
| 425 |
+
"logps/rejected": -119.75496673583984,
|
| 426 |
+
"loss": 0.6171,
|
| 427 |
+
"rewards/accuracies": 0.6625000238418579,
|
| 428 |
+
"rewards/chosen": -0.20113949477672577,
|
| 429 |
+
"rewards/margins": 0.19329218566417694,
|
| 430 |
+
"rewards/rejected": -0.3944316804409027,
|
| 431 |
+
"step": 280
|
| 432 |
+
},
|
| 433 |
+
{
|
| 434 |
+
"epoch": 0.29,
|
| 435 |
+
"learning_rate": 4.495956541020376e-06,
|
| 436 |
+
"logits/chosen": -2.3487744331359863,
|
| 437 |
+
"logits/rejected": -2.1971921920776367,
|
| 438 |
+
"logps/chosen": -150.26535034179688,
|
| 439 |
+
"logps/rejected": -133.6148681640625,
|
| 440 |
+
"loss": 0.6038,
|
| 441 |
+
"rewards/accuracies": 0.6625000238418579,
|
| 442 |
+
"rewards/chosen": -0.36308565735816956,
|
| 443 |
+
"rewards/margins": 0.2375185787677765,
|
| 444 |
+
"rewards/rejected": -0.6006041765213013,
|
| 445 |
+
"step": 290
|
| 446 |
+
},
|
| 447 |
+
{
|
| 448 |
+
"epoch": 0.3,
|
| 449 |
+
"learning_rate": 4.443045148759978e-06,
|
| 450 |
+
"logits/chosen": -2.3486955165863037,
|
| 451 |
+
"logits/rejected": -2.1962506771087646,
|
| 452 |
+
"logps/chosen": -159.32064819335938,
|
| 453 |
+
"logps/rejected": -126.43644714355469,
|
| 454 |
+
"loss": 0.5853,
|
| 455 |
+
"rewards/accuracies": 0.7250000238418579,
|
| 456 |
+
"rewards/chosen": -0.44245219230651855,
|
| 457 |
+
"rewards/margins": 0.2853606343269348,
|
| 458 |
+
"rewards/rejected": -0.7278127670288086,
|
| 459 |
+
"step": 300
|
| 460 |
+
},
|
| 461 |
+
{
|
| 462 |
+
"epoch": 0.3,
|
| 463 |
+
"learning_rate": 4.3878384140516025e-06,
|
| 464 |
+
"logits/chosen": -2.3736767768859863,
|
| 465 |
+
"logits/rejected": -2.249598741531372,
|
| 466 |
+
"logps/chosen": -146.68006896972656,
|
| 467 |
+
"logps/rejected": -135.25997924804688,
|
| 468 |
+
"loss": 0.5898,
|
| 469 |
+
"rewards/accuracies": 0.737500011920929,
|
| 470 |
+
"rewards/chosen": -0.41068369150161743,
|
| 471 |
+
"rewards/margins": 0.26723065972328186,
|
| 472 |
+
"rewards/rejected": -0.6779143214225769,
|
| 473 |
+
"step": 310
|
| 474 |
+
},
|
| 475 |
+
{
|
| 476 |
+
"epoch": 0.31,
|
| 477 |
+
"learning_rate": 4.330401553270522e-06,
|
| 478 |
+
"logits/chosen": -2.298905372619629,
|
| 479 |
+
"logits/rejected": -2.1949081420898438,
|
| 480 |
+
"logps/chosen": -146.1105194091797,
|
| 481 |
+
"logps/rejected": -136.11727905273438,
|
| 482 |
+
"loss": 0.6003,
|
| 483 |
+
"rewards/accuracies": 0.7124999761581421,
|
| 484 |
+
"rewards/chosen": -0.551005482673645,
|
| 485 |
+
"rewards/margins": 0.2657201588153839,
|
| 486 |
+
"rewards/rejected": -0.8167255520820618,
|
| 487 |
+
"step": 320
|
| 488 |
+
},
|
| 489 |
+
{
|
| 490 |
+
"epoch": 0.32,
|
| 491 |
+
"learning_rate": 4.2708024172665795e-06,
|
| 492 |
+
"logits/chosen": -2.402360439300537,
|
| 493 |
+
"logits/rejected": -2.2175159454345703,
|
| 494 |
+
"logps/chosen": -150.5807342529297,
|
| 495 |
+
"logps/rejected": -125.05631256103516,
|
| 496 |
+
"loss": 0.5465,
|
| 497 |
+
"rewards/accuracies": 0.824999988079071,
|
| 498 |
+
"rewards/chosen": -0.7037911415100098,
|
| 499 |
+
"rewards/margins": 0.3860488533973694,
|
| 500 |
+
"rewards/rejected": -1.089840054512024,
|
| 501 |
+
"step": 330
|
| 502 |
+
},
|
| 503 |
+
{
|
| 504 |
+
"epoch": 0.33,
|
| 505 |
+
"learning_rate": 4.209111411211174e-06,
|
| 506 |
+
"logits/chosen": -2.413839340209961,
|
| 507 |
+
"logits/rejected": -2.253542423248291,
|
| 508 |
+
"logps/chosen": -142.4720001220703,
|
| 509 |
+
"logps/rejected": -127.50254821777344,
|
| 510 |
+
"loss": 0.5395,
|
| 511 |
+
"rewards/accuracies": 0.75,
|
| 512 |
+
"rewards/chosen": -0.7327712178230286,
|
| 513 |
+
"rewards/margins": 0.4077509045600891,
|
| 514 |
+
"rewards/rejected": -1.1405221223831177,
|
| 515 |
+
"step": 340
|
| 516 |
+
},
|
| 517 |
+
{
|
| 518 |
+
"epoch": 0.34,
|
| 519 |
+
"learning_rate": 4.145401411426788e-06,
|
| 520 |
+
"logits/chosen": -2.4574408531188965,
|
| 521 |
+
"logits/rejected": -2.3272013664245605,
|
| 522 |
+
"logps/chosen": -143.1080322265625,
|
| 523 |
+
"logps/rejected": -130.01637268066406,
|
| 524 |
+
"loss": 0.5795,
|
| 525 |
+
"rewards/accuracies": 0.7250000238418579,
|
| 526 |
+
"rewards/chosen": -0.7288345098495483,
|
| 527 |
+
"rewards/margins": 0.36709967255592346,
|
| 528 |
+
"rewards/rejected": -1.0959341526031494,
|
| 529 |
+
"step": 350
|
| 530 |
+
},
|
| 531 |
+
{
|
| 532 |
+
"epoch": 0.35,
|
| 533 |
+
"learning_rate": 4.079747679297314e-06,
|
| 534 |
+
"logits/chosen": -2.295055627822876,
|
| 535 |
+
"logits/rejected": -2.179664134979248,
|
| 536 |
+
"logps/chosen": -157.7494354248047,
|
| 537 |
+
"logps/rejected": -137.73941040039062,
|
| 538 |
+
"loss": 0.5973,
|
| 539 |
+
"rewards/accuracies": 0.637499988079071,
|
| 540 |
+
"rewards/chosen": -0.8279803395271301,
|
| 541 |
+
"rewards/margins": 0.3018009066581726,
|
| 542 |
+
"rewards/rejected": -1.1297812461853027,
|
| 543 |
+
"step": 360
|
| 544 |
+
},
|
| 545 |
+
{
|
| 546 |
+
"epoch": 0.36,
|
| 547 |
+
"learning_rate": 4.012227772360889e-06,
|
| 548 |
+
"logits/chosen": -2.2948107719421387,
|
| 549 |
+
"logits/rejected": -2.1263232231140137,
|
| 550 |
+
"logps/chosen": -167.21339416503906,
|
| 551 |
+
"logps/rejected": -150.82669067382812,
|
| 552 |
+
"loss": 0.5349,
|
| 553 |
+
"rewards/accuracies": 0.7875000238418579,
|
| 554 |
+
"rewards/chosen": -0.9286147356033325,
|
| 555 |
+
"rewards/margins": 0.44435009360313416,
|
| 556 |
+
"rewards/rejected": -1.3729647397994995,
|
| 557 |
+
"step": 370
|
| 558 |
+
},
|
| 559 |
+
{
|
| 560 |
+
"epoch": 0.37,
|
| 561 |
+
"learning_rate": 3.942921452690245e-06,
|
| 562 |
+
"logits/chosen": -2.3513195514678955,
|
| 563 |
+
"logits/rejected": -2.25927734375,
|
| 564 |
+
"logps/chosen": -173.61428833007812,
|
| 565 |
+
"logps/rejected": -163.9345703125,
|
| 566 |
+
"loss": 0.6164,
|
| 567 |
+
"rewards/accuracies": 0.7124999761581421,
|
| 568 |
+
"rewards/chosen": -1.3186113834381104,
|
| 569 |
+
"rewards/margins": 0.24015231430530548,
|
| 570 |
+
"rewards/rejected": -1.558763861656189,
|
| 571 |
+
"step": 380
|
| 572 |
+
},
|
| 573 |
+
{
|
| 574 |
+
"epoch": 0.38,
|
| 575 |
+
"learning_rate": 3.871910592668817e-06,
|
| 576 |
+
"logits/chosen": -2.433640480041504,
|
| 577 |
+
"logits/rejected": -2.300931215286255,
|
| 578 |
+
"logps/chosen": -170.8921356201172,
|
| 579 |
+
"logps/rejected": -165.83462524414062,
|
| 580 |
+
"loss": 0.503,
|
| 581 |
+
"rewards/accuracies": 0.8125,
|
| 582 |
+
"rewards/chosen": -1.420196294784546,
|
| 583 |
+
"rewards/margins": 0.5389910340309143,
|
| 584 |
+
"rewards/rejected": -1.9591872692108154,
|
| 585 |
+
"step": 390
|
| 586 |
+
},
|
| 587 |
+
{
|
| 588 |
+
"epoch": 0.39,
|
| 589 |
+
"learning_rate": 3.799279078273921e-06,
|
| 590 |
+
"logits/chosen": -2.3466134071350098,
|
| 591 |
+
"logits/rejected": -2.157196044921875,
|
| 592 |
+
"logps/chosen": -177.853271484375,
|
| 593 |
+
"logps/rejected": -158.73776245117188,
|
| 594 |
+
"loss": 0.4905,
|
| 595 |
+
"rewards/accuracies": 0.762499988079071,
|
| 596 |
+
"rewards/chosen": -1.5086350440979004,
|
| 597 |
+
"rewards/margins": 0.6321147680282593,
|
| 598 |
+
"rewards/rejected": -2.140749454498291,
|
| 599 |
+
"step": 400
|
| 600 |
+
},
|
| 601 |
+
{
|
| 602 |
+
"epoch": 0.39,
|
| 603 |
+
"eval_logits/chosen": -2.195096015930176,
|
| 604 |
+
"eval_logits/rejected": -2.0950398445129395,
|
| 605 |
+
"eval_logps/chosen": -177.19459533691406,
|
| 606 |
+
"eval_logps/rejected": -168.9884490966797,
|
| 607 |
+
"eval_loss": 0.5550708174705505,
|
| 608 |
+
"eval_rewards/accuracies": 0.6957547068595886,
|
| 609 |
+
"eval_rewards/chosen": -1.6581227779388428,
|
| 610 |
+
"eval_rewards/margins": 0.4946018159389496,
|
| 611 |
+
"eval_rewards/rejected": -2.152724504470825,
|
| 612 |
+
"eval_runtime": 417.866,
|
| 613 |
+
"eval_samples_per_second": 1.0,
|
| 614 |
+
"eval_steps_per_second": 0.127,
|
| 615 |
+
"step": 400
|
| 616 |
+
},
|
| 617 |
+
{
|
| 618 |
+
"epoch": 0.4,
|
| 619 |
+
"learning_rate": 3.725112709981249e-06,
|
| 620 |
+
"logits/chosen": -2.20538592338562,
|
| 621 |
+
"logits/rejected": -2.059528112411499,
|
| 622 |
+
"logps/chosen": -192.91726684570312,
|
| 623 |
+
"logps/rejected": -185.90341186523438,
|
| 624 |
+
"loss": 0.5449,
|
| 625 |
+
"rewards/accuracies": 0.737500011920929,
|
| 626 |
+
"rewards/chosen": -1.8029924631118774,
|
| 627 |
+
"rewards/margins": 0.5775827765464783,
|
| 628 |
+
"rewards/rejected": -2.38057541847229,
|
| 629 |
+
"step": 410
|
| 630 |
+
},
|
| 631 |
+
{
|
| 632 |
+
"epoch": 0.41,
|
| 633 |
+
"learning_rate": 3.649499101407737e-06,
|
| 634 |
+
"logits/chosen": -2.363370895385742,
|
| 635 |
+
"logits/rejected": -2.2048511505126953,
|
| 636 |
+
"logps/chosen": -224.8810577392578,
|
| 637 |
+
"logps/rejected": -214.8511505126953,
|
| 638 |
+
"loss": 0.5825,
|
| 639 |
+
"rewards/accuracies": 0.6875,
|
| 640 |
+
"rewards/chosen": -2.7866973876953125,
|
| 641 |
+
"rewards/margins": 0.5168424844741821,
|
| 642 |
+
"rewards/rejected": -3.303539991378784,
|
| 643 |
+
"step": 420
|
| 644 |
+
},
|
| 645 |
+
{
|
| 646 |
+
"epoch": 0.42,
|
| 647 |
+
"learning_rate": 3.5725275758125564e-06,
|
| 648 |
+
"logits/chosen": -2.27677059173584,
|
| 649 |
+
"logits/rejected": -2.1246485710144043,
|
| 650 |
+
"logps/chosen": -200.77066040039062,
|
| 651 |
+
"logps/rejected": -190.68533325195312,
|
| 652 |
+
"loss": 0.4805,
|
| 653 |
+
"rewards/accuracies": 0.800000011920929,
|
| 654 |
+
"rewards/chosen": -2.6391139030456543,
|
| 655 |
+
"rewards/margins": 0.7239618897438049,
|
| 656 |
+
"rewards/rejected": -3.3630757331848145,
|
| 657 |
+
"step": 430
|
| 658 |
+
},
|
| 659 |
+
{
|
| 660 |
+
"epoch": 0.43,
|
| 661 |
+
"learning_rate": 3.494289060578478e-06,
|
| 662 |
+
"logits/chosen": -2.3822944164276123,
|
| 663 |
+
"logits/rejected": -2.2915594577789307,
|
| 664 |
+
"logps/chosen": -169.66128540039062,
|
| 665 |
+
"logps/rejected": -174.19607543945312,
|
| 666 |
+
"loss": 0.4812,
|
| 667 |
+
"rewards/accuracies": 0.699999988079071,
|
| 668 |
+
"rewards/chosen": -1.6458194255828857,
|
| 669 |
+
"rewards/margins": 0.7625138163566589,
|
| 670 |
+
"rewards/rejected": -2.4083335399627686,
|
| 671 |
+
"step": 440
|
| 672 |
+
},
|
| 673 |
+
{
|
| 674 |
+
"epoch": 0.44,
|
| 675 |
+
"learning_rate": 3.414875979798272e-06,
|
| 676 |
+
"logits/chosen": -2.327730178833008,
|
| 677 |
+
"logits/rejected": -2.2062036991119385,
|
| 678 |
+
"logps/chosen": -194.1064453125,
|
| 679 |
+
"logps/rejected": -193.63107299804688,
|
| 680 |
+
"loss": 0.4394,
|
| 681 |
+
"rewards/accuracies": 0.8374999761581421,
|
| 682 |
+
"rewards/chosen": -1.4300081729888916,
|
| 683 |
+
"rewards/margins": 0.892192006111145,
|
| 684 |
+
"rewards/rejected": -2.322200059890747,
|
| 685 |
+
"step": 450
|
| 686 |
+
},
|
| 687 |
+
{
|
| 688 |
+
"epoch": 0.45,
|
| 689 |
+
"learning_rate": 3.3343821450930196e-06,
|
| 690 |
+
"logits/chosen": -2.454336166381836,
|
| 691 |
+
"logits/rejected": -2.2937960624694824,
|
| 692 |
+
"logps/chosen": -201.33657836914062,
|
| 693 |
+
"logps/rejected": -188.39266967773438,
|
| 694 |
+
"loss": 0.4758,
|
| 695 |
+
"rewards/accuracies": 0.75,
|
| 696 |
+
"rewards/chosen": -2.0642800331115723,
|
| 697 |
+
"rewards/margins": 0.8783187866210938,
|
| 698 |
+
"rewards/rejected": -2.942598819732666,
|
| 699 |
+
"step": 460
|
| 700 |
+
},
|
| 701 |
+
{
|
| 702 |
+
"epoch": 0.46,
|
| 703 |
+
"learning_rate": 3.252902644791325e-06,
|
| 704 |
+
"logits/chosen": -2.188424587249756,
|
| 705 |
+
"logits/rejected": -2.041105031967163,
|
| 706 |
+
"logps/chosen": -214.7936248779297,
|
| 707 |
+
"logps/rejected": -222.3824462890625,
|
| 708 |
+
"loss": 0.4536,
|
| 709 |
+
"rewards/accuracies": 0.800000011920929,
|
| 710 |
+
"rewards/chosen": -2.7383971214294434,
|
| 711 |
+
"rewards/margins": 0.9385073781013489,
|
| 712 |
+
"rewards/rejected": -3.6769042015075684,
|
| 713 |
+
"step": 470
|
| 714 |
+
},
|
| 715 |
+
{
|
| 716 |
+
"epoch": 0.47,
|
| 717 |
+
"learning_rate": 3.170533731600339e-06,
|
| 718 |
+
"logits/chosen": -2.3067922592163086,
|
| 719 |
+
"logits/rejected": -2.1540229320526123,
|
| 720 |
+
"logps/chosen": -204.9595184326172,
|
| 721 |
+
"logps/rejected": -204.6565704345703,
|
| 722 |
+
"loss": 0.4887,
|
| 723 |
+
"rewards/accuracies": 0.75,
|
| 724 |
+
"rewards/chosen": -2.3199477195739746,
|
| 725 |
+
"rewards/margins": 0.8953973650932312,
|
| 726 |
+
"rewards/rejected": -3.2153449058532715,
|
| 727 |
+
"step": 480
|
| 728 |
+
},
|
| 729 |
+
{
|
| 730 |
+
"epoch": 0.48,
|
| 731 |
+
"learning_rate": 3.0873727089012816e-06,
|
| 732 |
+
"logits/chosen": -2.4507811069488525,
|
| 733 |
+
"logits/rejected": -2.357919931411743,
|
| 734 |
+
"logps/chosen": -212.8636932373047,
|
| 735 |
+
"logps/rejected": -208.5207977294922,
|
| 736 |
+
"loss": 0.5393,
|
| 737 |
+
"rewards/accuracies": 0.75,
|
| 738 |
+
"rewards/chosen": -2.4198460578918457,
|
| 739 |
+
"rewards/margins": 0.7457529306411743,
|
| 740 |
+
"rewards/rejected": -3.1655986309051514,
|
| 741 |
+
"step": 490
|
| 742 |
+
},
|
| 743 |
+
{
|
| 744 |
+
"epoch": 0.49,
|
| 745 |
+
"learning_rate": 3.0035178158038026e-06,
|
| 746 |
+
"logits/chosen": -2.2298638820648193,
|
| 747 |
+
"logits/rejected": -2.035947322845459,
|
| 748 |
+
"logps/chosen": -217.26846313476562,
|
| 749 |
+
"logps/rejected": -209.74526977539062,
|
| 750 |
+
"loss": 0.4886,
|
| 751 |
+
"rewards/accuracies": 0.7250000238418579,
|
| 752 |
+
"rewards/chosen": -2.4076380729675293,
|
| 753 |
+
"rewards/margins": 0.9488040208816528,
|
| 754 |
+
"rewards/rejected": -3.3564422130584717,
|
| 755 |
+
"step": 500
|
| 756 |
+
},
|
| 757 |
+
{
|
| 758 |
+
"epoch": 0.5,
|
| 759 |
+
"learning_rate": 2.919068111094937e-06,
|
| 760 |
+
"logits/chosen": -2.3187146186828613,
|
| 761 |
+
"logits/rejected": -2.193861484527588,
|
| 762 |
+
"logps/chosen": -185.0901641845703,
|
| 763 |
+
"logps/rejected": -195.90634155273438,
|
| 764 |
+
"loss": 0.552,
|
| 765 |
+
"rewards/accuracies": 0.6875,
|
| 766 |
+
"rewards/chosen": -2.132998466491699,
|
| 767 |
+
"rewards/margins": 0.827830970287323,
|
| 768 |
+
"rewards/rejected": -2.960829257965088,
|
| 769 |
+
"step": 510
|
| 770 |
+
},
|
| 771 |
+
{
|
| 772 |
+
"epoch": 0.51,
|
| 773 |
+
"learning_rate": 2.8341233562197895e-06,
|
| 774 |
+
"logits/chosen": -2.3116376399993896,
|
| 775 |
+
"logits/rejected": -2.246950626373291,
|
| 776 |
+
"logps/chosen": -172.0282440185547,
|
| 777 |
+
"logps/rejected": -177.47348022460938,
|
| 778 |
+
"loss": 0.5152,
|
| 779 |
+
"rewards/accuracies": 0.762499988079071,
|
| 780 |
+
"rewards/chosen": -1.7447055578231812,
|
| 781 |
+
"rewards/margins": 0.754051685333252,
|
| 782 |
+
"rewards/rejected": -2.4987568855285645,
|
| 783 |
+
"step": 520
|
| 784 |
+
},
|
| 785 |
+
{
|
| 786 |
+
"epoch": 0.52,
|
| 787 |
+
"learning_rate": 2.7487838974321352e-06,
|
| 788 |
+
"logits/chosen": -2.2577805519104004,
|
| 789 |
+
"logits/rejected": -2.143658399581909,
|
| 790 |
+
"logps/chosen": -178.53317260742188,
|
| 791 |
+
"logps/rejected": -182.5507049560547,
|
| 792 |
+
"loss": 0.4719,
|
| 793 |
+
"rewards/accuracies": 0.800000011920929,
|
| 794 |
+
"rewards/chosen": -1.789046287536621,
|
| 795 |
+
"rewards/margins": 0.8771921396255493,
|
| 796 |
+
"rewards/rejected": -2.666238307952881,
|
| 797 |
+
"step": 530
|
| 798 |
+
},
|
| 799 |
+
{
|
| 800 |
+
"epoch": 0.53,
|
| 801 |
+
"learning_rate": 2.6631505472541997e-06,
|
| 802 |
+
"logits/chosen": -2.2621216773986816,
|
| 803 |
+
"logits/rejected": -2.1167335510253906,
|
| 804 |
+
"logps/chosen": -193.0259246826172,
|
| 805 |
+
"logps/rejected": -198.51705932617188,
|
| 806 |
+
"loss": 0.4442,
|
| 807 |
+
"rewards/accuracies": 0.762499988079071,
|
| 808 |
+
"rewards/chosen": -1.8945804834365845,
|
| 809 |
+
"rewards/margins": 1.0066120624542236,
|
| 810 |
+
"rewards/rejected": -2.9011926651000977,
|
| 811 |
+
"step": 540
|
| 812 |
+
},
|
| 813 |
+
{
|
| 814 |
+
"epoch": 0.54,
|
| 815 |
+
"learning_rate": 2.5773244653856173e-06,
|
| 816 |
+
"logits/chosen": -2.2354609966278076,
|
| 817 |
+
"logits/rejected": -2.1045310497283936,
|
| 818 |
+
"logps/chosen": -205.27163696289062,
|
| 819 |
+
"logps/rejected": -207.8455810546875,
|
| 820 |
+
"loss": 0.4791,
|
| 821 |
+
"rewards/accuracies": 0.7250000238418579,
|
| 822 |
+
"rewards/chosen": -2.161865711212158,
|
| 823 |
+
"rewards/margins": 0.9317368268966675,
|
| 824 |
+
"rewards/rejected": -3.0936026573181152,
|
| 825 |
+
"step": 550
|
| 826 |
+
},
|
| 827 |
+
{
|
| 828 |
+
"epoch": 0.55,
|
| 829 |
+
"learning_rate": 2.4914070392022717e-06,
|
| 830 |
+
"logits/chosen": -2.274534225463867,
|
| 831 |
+
"logits/rejected": -2.158811569213867,
|
| 832 |
+
"logps/chosen": -208.08349609375,
|
| 833 |
+
"logps/rejected": -211.2762451171875,
|
| 834 |
+
"loss": 0.4978,
|
| 835 |
+
"rewards/accuracies": 0.75,
|
| 836 |
+
"rewards/chosen": -2.0876078605651855,
|
| 837 |
+
"rewards/margins": 0.8888559341430664,
|
| 838 |
+
"rewards/rejected": -2.976463794708252,
|
| 839 |
+
"step": 560
|
| 840 |
+
},
|
| 841 |
+
{
|
| 842 |
+
"epoch": 0.56,
|
| 843 |
+
"learning_rate": 2.4054997639861778e-06,
|
| 844 |
+
"logits/chosen": -2.1874241828918457,
|
| 845 |
+
"logits/rejected": -2.0126781463623047,
|
| 846 |
+
"logps/chosen": -207.1369171142578,
|
| 847 |
+
"logps/rejected": -207.59768676757812,
|
| 848 |
+
"loss": 0.4732,
|
| 849 |
+
"rewards/accuracies": 0.800000011920929,
|
| 850 |
+
"rewards/chosen": -2.0929489135742188,
|
| 851 |
+
"rewards/margins": 1.14609694480896,
|
| 852 |
+
"rewards/rejected": -3.2390456199645996,
|
| 853 |
+
"step": 570
|
| 854 |
+
},
|
| 855 |
+
{
|
| 856 |
+
"epoch": 0.57,
|
| 857 |
+
"learning_rate": 2.3197041230278905e-06,
|
| 858 |
+
"logits/chosen": -2.3066487312316895,
|
| 859 |
+
"logits/rejected": -2.18741512298584,
|
| 860 |
+
"logps/chosen": -204.58642578125,
|
| 861 |
+
"logps/rejected": -222.44973754882812,
|
| 862 |
+
"loss": 0.427,
|
| 863 |
+
"rewards/accuracies": 0.8125,
|
| 864 |
+
"rewards/chosen": -2.2493700981140137,
|
| 865 |
+
"rewards/margins": 1.2541824579238892,
|
| 866 |
+
"rewards/rejected": -3.5035526752471924,
|
| 867 |
+
"step": 580
|
| 868 |
+
},
|
| 869 |
+
{
|
| 870 |
+
"epoch": 0.58,
|
| 871 |
+
"learning_rate": 2.234121467743082e-06,
|
| 872 |
+
"logits/chosen": -2.3349661827087402,
|
| 873 |
+
"logits/rejected": -2.252894163131714,
|
| 874 |
+
"logps/chosen": -208.3444061279297,
|
| 875 |
+
"logps/rejected": -211.37295532226562,
|
| 876 |
+
"loss": 0.5968,
|
| 877 |
+
"rewards/accuracies": 0.7250000238418579,
|
| 878 |
+
"rewards/chosen": -2.635042905807495,
|
| 879 |
+
"rewards/margins": 0.8392454981803894,
|
| 880 |
+
"rewards/rejected": -3.4742884635925293,
|
| 881 |
+
"step": 590
|
| 882 |
+
},
|
| 883 |
+
{
|
| 884 |
+
"epoch": 0.59,
|
| 885 |
+
"learning_rate": 2.148852897944905e-06,
|
| 886 |
+
"logits/chosen": -2.3977198600769043,
|
| 887 |
+
"logits/rejected": -2.2428812980651855,
|
| 888 |
+
"logps/chosen": -208.42538452148438,
|
| 889 |
+
"logps/rejected": -219.8905487060547,
|
| 890 |
+
"loss": 0.4249,
|
| 891 |
+
"rewards/accuracies": 0.8374999761581421,
|
| 892 |
+
"rewards/chosen": -2.645089626312256,
|
| 893 |
+
"rewards/margins": 1.1129385232925415,
|
| 894 |
+
"rewards/rejected": -3.758028030395508,
|
| 895 |
+
"step": 600
|
| 896 |
+
},
|
| 897 |
+
{
|
| 898 |
+
"epoch": 0.59,
|
| 899 |
+
"eval_logits/chosen": -2.17734956741333,
|
| 900 |
+
"eval_logits/rejected": -2.078249454498291,
|
| 901 |
+
"eval_logps/chosen": -237.10446166992188,
|
| 902 |
+
"eval_logps/rejected": -241.38665771484375,
|
| 903 |
+
"eval_loss": 0.5327094793319702,
|
| 904 |
+
"eval_rewards/accuracies": 0.724056601524353,
|
| 905 |
+
"eval_rewards/chosen": -3.455418348312378,
|
| 906 |
+
"eval_rewards/margins": 0.8692519068717957,
|
| 907 |
+
"eval_rewards/rejected": -4.324670314788818,
|
| 908 |
+
"eval_runtime": 423.7708,
|
| 909 |
+
"eval_samples_per_second": 0.986,
|
| 910 |
+
"eval_steps_per_second": 0.125,
|
| 911 |
+
"step": 600
|
| 912 |
+
},
|
| 913 |
+
{
|
| 914 |
+
"epoch": 0.6,
|
| 915 |
+
"learning_rate": 2.063999142413574e-06,
|
| 916 |
+
"logits/chosen": -2.2915334701538086,
|
| 917 |
+
"logits/rejected": -2.1494946479797363,
|
| 918 |
+
"logps/chosen": -247.0482177734375,
|
| 919 |
+
"logps/rejected": -239.52462768554688,
|
| 920 |
+
"loss": 0.489,
|
| 921 |
+
"rewards/accuracies": 0.762499988079071,
|
| 922 |
+
"rewards/chosen": -3.5384364128112793,
|
| 923 |
+
"rewards/margins": 0.9590060114860535,
|
| 924 |
+
"rewards/rejected": -4.497443199157715,
|
| 925 |
+
"step": 610
|
| 926 |
+
},
|
| 927 |
+
{
|
| 928 |
+
"epoch": 0.61,
|
| 929 |
+
"learning_rate": 1.9796604399042547e-06,
|
| 930 |
+
"logits/chosen": -2.3757712841033936,
|
| 931 |
+
"logits/rejected": -2.2203211784362793,
|
| 932 |
+
"logps/chosen": -268.9779052734375,
|
| 933 |
+
"logps/rejected": -274.21771240234375,
|
| 934 |
+
"loss": 0.4595,
|
| 935 |
+
"rewards/accuracies": 0.7875000238418579,
|
| 936 |
+
"rewards/chosen": -4.319927215576172,
|
| 937 |
+
"rewards/margins": 1.1538012027740479,
|
| 938 |
+
"rewards/rejected": -5.473728179931641,
|
| 939 |
+
"step": 620
|
| 940 |
+
},
|
| 941 |
+
{
|
| 942 |
+
"epoch": 0.62,
|
| 943 |
+
"learning_rate": 1.8959364207338216e-06,
|
| 944 |
+
"logits/chosen": -2.345416784286499,
|
| 945 |
+
"logits/rejected": -2.180387020111084,
|
| 946 |
+
"logps/chosen": -245.0520477294922,
|
| 947 |
+
"logps/rejected": -258.79498291015625,
|
| 948 |
+
"loss": 0.4835,
|
| 949 |
+
"rewards/accuracies": 0.737500011920929,
|
| 950 |
+
"rewards/chosen": -4.156452655792236,
|
| 951 |
+
"rewards/margins": 1.1458499431610107,
|
| 952 |
+
"rewards/rejected": -5.302302360534668,
|
| 953 |
+
"step": 630
|
| 954 |
+
},
|
| 955 |
+
{
|
| 956 |
+
"epoch": 0.63,
|
| 957 |
+
"learning_rate": 1.8129259890863825e-06,
|
| 958 |
+
"logits/chosen": -2.3113701343536377,
|
| 959 |
+
"logits/rejected": -2.200329303741455,
|
| 960 |
+
"logps/chosen": -258.2785949707031,
|
| 961 |
+
"logps/rejected": -273.8367614746094,
|
| 962 |
+
"loss": 0.5494,
|
| 963 |
+
"rewards/accuracies": 0.675000011920929,
|
| 964 |
+
"rewards/chosen": -3.8280506134033203,
|
| 965 |
+
"rewards/margins": 1.0454813241958618,
|
| 966 |
+
"rewards/rejected": -4.873531818389893,
|
| 967 |
+
"step": 640
|
| 968 |
+
},
|
| 969 |
+
{
|
| 970 |
+
"epoch": 0.64,
|
| 971 |
+
"learning_rate": 1.7307272061765738e-06,
|
| 972 |
+
"logits/chosen": -2.332291841506958,
|
| 973 |
+
"logits/rejected": -2.238374948501587,
|
| 974 |
+
"logps/chosen": -245.2693328857422,
|
| 975 |
+
"logps/rejected": -258.13922119140625,
|
| 976 |
+
"loss": 0.4966,
|
| 977 |
+
"rewards/accuracies": 0.762499988079071,
|
| 978 |
+
"rewards/chosen": -3.4873454570770264,
|
| 979 |
+
"rewards/margins": 1.1437709331512451,
|
| 980 |
+
"rewards/rejected": -4.6311163902282715,
|
| 981 |
+
"step": 650
|
| 982 |
+
},
|
| 983 |
+
{
|
| 984 |
+
"epoch": 0.65,
|
| 985 |
+
"learning_rate": 1.649437174408685e-06,
|
| 986 |
+
"logits/chosen": -2.2653393745422363,
|
| 987 |
+
"logits/rejected": -2.1328892707824707,
|
| 988 |
+
"logps/chosen": -225.3964385986328,
|
| 989 |
+
"logps/rejected": -243.22384643554688,
|
| 990 |
+
"loss": 0.5442,
|
| 991 |
+
"rewards/accuracies": 0.75,
|
| 992 |
+
"rewards/chosen": -2.7976624965667725,
|
| 993 |
+
"rewards/margins": 1.2127068042755127,
|
| 994 |
+
"rewards/rejected": -4.010369300842285,
|
| 995 |
+
"step": 660
|
| 996 |
+
},
|
| 997 |
+
{
|
| 998 |
+
"epoch": 0.66,
|
| 999 |
+
"learning_rate": 1.569151922668422e-06,
|
| 1000 |
+
"logits/chosen": -2.367928981781006,
|
| 1001 |
+
"logits/rejected": -2.253605365753174,
|
| 1002 |
+
"logps/chosen": -201.37025451660156,
|
| 1003 |
+
"logps/rejected": -218.9671630859375,
|
| 1004 |
+
"loss": 0.5362,
|
| 1005 |
+
"rewards/accuracies": 0.7749999761581421,
|
| 1006 |
+
"rewards/chosen": -2.52730131149292,
|
| 1007 |
+
"rewards/margins": 1.1274640560150146,
|
| 1008 |
+
"rewards/rejected": -3.6547648906707764,
|
| 1009 |
+
"step": 670
|
| 1010 |
+
},
|
| 1011 |
+
{
|
| 1012 |
+
"epoch": 0.67,
|
| 1013 |
+
"learning_rate": 1.4899662928828428e-06,
|
| 1014 |
+
"logits/chosen": -2.309407949447632,
|
| 1015 |
+
"logits/rejected": -2.1246845722198486,
|
| 1016 |
+
"logps/chosen": -186.25376892089844,
|
| 1017 |
+
"logps/rejected": -205.3202667236328,
|
| 1018 |
+
"loss": 0.4534,
|
| 1019 |
+
"rewards/accuracies": 0.8125,
|
| 1020 |
+
"rewards/chosen": -2.198615550994873,
|
| 1021 |
+
"rewards/margins": 1.3432915210723877,
|
| 1022 |
+
"rewards/rejected": -3.5419068336486816,
|
| 1023 |
+
"step": 680
|
| 1024 |
+
},
|
| 1025 |
+
{
|
| 1026 |
+
"epoch": 0.68,
|
| 1027 |
+
"learning_rate": 1.4119738279824507e-06,
|
| 1028 |
+
"logits/chosen": -2.233764886856079,
|
| 1029 |
+
"logits/rejected": -2.0911850929260254,
|
| 1030 |
+
"logps/chosen": -194.0051727294922,
|
| 1031 |
+
"logps/rejected": -202.19174194335938,
|
| 1032 |
+
"loss": 0.5816,
|
| 1033 |
+
"rewards/accuracies": 0.7250000238418579,
|
| 1034 |
+
"rewards/chosen": -2.272157907485962,
|
| 1035 |
+
"rewards/margins": 0.9171239137649536,
|
| 1036 |
+
"rewards/rejected": -3.189281463623047,
|
| 1037 |
+
"step": 690
|
| 1038 |
+
},
|
| 1039 |
+
{
|
| 1040 |
+
"epoch": 0.69,
|
| 1041 |
+
"learning_rate": 1.3352666613978152e-06,
|
| 1042 |
+
"logits/chosen": -2.1566410064697266,
|
| 1043 |
+
"logits/rejected": -2.075209379196167,
|
| 1044 |
+
"logps/chosen": -198.47071838378906,
|
| 1045 |
+
"logps/rejected": -202.29379272460938,
|
| 1046 |
+
"loss": 0.5355,
|
| 1047 |
+
"rewards/accuracies": 0.7250000238418579,
|
| 1048 |
+
"rewards/chosen": -2.325533151626587,
|
| 1049 |
+
"rewards/margins": 0.8646724820137024,
|
| 1050 |
+
"rewards/rejected": -3.1902058124542236,
|
| 1051 |
+
"step": 700
|
| 1052 |
+
},
|
| 1053 |
+
{
|
| 1054 |
+
"epoch": 0.7,
|
| 1055 |
+
"learning_rate": 1.2599354082212523e-06,
|
| 1056 |
+
"logits/chosen": -2.222766876220703,
|
| 1057 |
+
"logits/rejected": -2.0928432941436768,
|
| 1058 |
+
"logps/chosen": -190.0797119140625,
|
| 1059 |
+
"logps/rejected": -198.69985961914062,
|
| 1060 |
+
"loss": 0.5152,
|
| 1061 |
+
"rewards/accuracies": 0.737500011920929,
|
| 1062 |
+
"rewards/chosen": -2.1240830421447754,
|
| 1063 |
+
"rewards/margins": 0.9723888635635376,
|
| 1064 |
+
"rewards/rejected": -3.0964715480804443,
|
| 1065 |
+
"step": 710
|
| 1066 |
+
},
|
| 1067 |
+
{
|
| 1068 |
+
"epoch": 0.71,
|
| 1069 |
+
"learning_rate": 1.186069058162127e-06,
|
| 1070 |
+
"logits/chosen": -2.3559908866882324,
|
| 1071 |
+
"logits/rejected": -2.249436140060425,
|
| 1072 |
+
"logps/chosen": -197.04818725585938,
|
| 1073 |
+
"logps/rejected": -201.38693237304688,
|
| 1074 |
+
"loss": 0.5537,
|
| 1075 |
+
"rewards/accuracies": 0.737500011920929,
|
| 1076 |
+
"rewards/chosen": -2.438502788543701,
|
| 1077 |
+
"rewards/margins": 0.8104500770568848,
|
| 1078 |
+
"rewards/rejected": -3.248952865600586,
|
| 1079 |
+
"step": 720
|
| 1080 |
+
},
|
| 1081 |
+
{
|
| 1082 |
+
"epoch": 0.72,
|
| 1083 |
+
"learning_rate": 1.113754870422254e-06,
|
| 1084 |
+
"logits/chosen": -2.40106201171875,
|
| 1085 |
+
"logits/rejected": -2.2911598682403564,
|
| 1086 |
+
"logps/chosen": -198.04409790039062,
|
| 1087 |
+
"logps/rejected": -214.9007568359375,
|
| 1088 |
+
"loss": 0.5454,
|
| 1089 |
+
"rewards/accuracies": 0.75,
|
| 1090 |
+
"rewards/chosen": -2.3586318492889404,
|
| 1091 |
+
"rewards/margins": 0.8204809427261353,
|
| 1092 |
+
"rewards/rejected": -3.179112672805786,
|
| 1093 |
+
"step": 730
|
| 1094 |
+
},
|
| 1095 |
+
{
|
| 1096 |
+
"epoch": 0.73,
|
| 1097 |
+
"learning_rate": 1.0430782706155545e-06,
|
| 1098 |
+
"logits/chosen": -2.4164676666259766,
|
| 1099 |
+
"logits/rejected": -2.2673213481903076,
|
| 1100 |
+
"logps/chosen": -204.253173828125,
|
| 1101 |
+
"logps/rejected": -209.43753051757812,
|
| 1102 |
+
"loss": 0.45,
|
| 1103 |
+
"rewards/accuracies": 0.7250000238418579,
|
| 1104 |
+
"rewards/chosen": -2.2196271419525146,
|
| 1105 |
+
"rewards/margins": 1.149065613746643,
|
| 1106 |
+
"rewards/rejected": -3.3686928749084473,
|
| 1107 |
+
"step": 740
|
| 1108 |
+
},
|
| 1109 |
+
{
|
| 1110 |
+
"epoch": 0.74,
|
| 1111 |
+
"learning_rate": 9.741227498537615e-07,
|
| 1112 |
+
"logits/chosen": -2.457432270050049,
|
| 1113 |
+
"logits/rejected": -2.321898937225342,
|
| 1114 |
+
"logps/chosen": -194.0103302001953,
|
| 1115 |
+
"logps/rejected": -204.46145629882812,
|
| 1116 |
+
"loss": 0.4596,
|
| 1117 |
+
"rewards/accuracies": 0.7749999761581421,
|
| 1118 |
+
"rewards/chosen": -2.4977972507476807,
|
| 1119 |
+
"rewards/margins": 1.0296275615692139,
|
| 1120 |
+
"rewards/rejected": -3.5274243354797363,
|
| 1121 |
+
"step": 750
|
| 1122 |
+
},
|
| 1123 |
+
{
|
| 1124 |
+
"epoch": 0.75,
|
| 1125 |
+
"learning_rate": 9.069697661173668e-07,
|
| 1126 |
+
"logits/chosen": -2.3087573051452637,
|
| 1127 |
+
"logits/rejected": -2.210942268371582,
|
| 1128 |
+
"logps/chosen": -205.71200561523438,
|
| 1129 |
+
"logps/rejected": -217.53921508789062,
|
| 1130 |
+
"loss": 0.6425,
|
| 1131 |
+
"rewards/accuracies": 0.6875,
|
| 1132 |
+
"rewards/chosen": -2.667379379272461,
|
| 1133 |
+
"rewards/margins": 0.8116022348403931,
|
| 1134 |
+
"rewards/rejected": -3.4789810180664062,
|
| 1135 |
+
"step": 760
|
| 1136 |
+
},
|
| 1137 |
+
{
|
| 1138 |
+
"epoch": 0.76,
|
| 1139 |
+
"learning_rate": 8.416986480283434e-07,
|
| 1140 |
+
"logits/chosen": -2.2393643856048584,
|
| 1141 |
+
"logits/rejected": -2.096648693084717,
|
| 1142 |
+
"logps/chosen": -205.7020721435547,
|
| 1143 |
+
"logps/rejected": -220.5606689453125,
|
| 1144 |
+
"loss": 0.5619,
|
| 1145 |
+
"rewards/accuracies": 0.699999988079071,
|
| 1146 |
+
"rewards/chosen": -2.711803913116455,
|
| 1147 |
+
"rewards/margins": 1.0723652839660645,
|
| 1148 |
+
"rewards/rejected": -3.7841694355010986,
|
| 1149 |
+
"step": 770
|
| 1150 |
+
},
|
| 1151 |
+
{
|
| 1152 |
+
"epoch": 0.77,
|
| 1153 |
+
"learning_rate": 7.783865011382876e-07,
|
| 1154 |
+
"logits/chosen": -2.2971372604370117,
|
| 1155 |
+
"logits/rejected": -2.1558918952941895,
|
| 1156 |
+
"logps/chosen": -213.41958618164062,
|
| 1157 |
+
"logps/rejected": -214.00326538085938,
|
| 1158 |
+
"loss": 0.5043,
|
| 1159 |
+
"rewards/accuracies": 0.699999988079071,
|
| 1160 |
+
"rewards/chosen": -2.363002300262451,
|
| 1161 |
+
"rewards/margins": 1.088847279548645,
|
| 1162 |
+
"rewards/rejected": -3.4518496990203857,
|
| 1163 |
+
"step": 780
|
| 1164 |
+
},
|
| 1165 |
+
{
|
| 1166 |
+
"epoch": 0.78,
|
| 1167 |
+
"learning_rate": 7.171081168427205e-07,
|
| 1168 |
+
"logits/chosen": -2.331343173980713,
|
| 1169 |
+
"logits/rejected": -2.165480613708496,
|
| 1170 |
+
"logps/chosen": -221.4240264892578,
|
| 1171 |
+
"logps/rejected": -226.91650390625,
|
| 1172 |
+
"loss": 0.473,
|
| 1173 |
+
"rewards/accuracies": 0.762499988079071,
|
| 1174 |
+
"rewards/chosen": -2.2688238620758057,
|
| 1175 |
+
"rewards/margins": 1.2109057903289795,
|
| 1176 |
+
"rewards/rejected": -3.479729413986206,
|
| 1177 |
+
"step": 790
|
| 1178 |
+
},
|
| 1179 |
+
{
|
| 1180 |
+
"epoch": 0.79,
|
| 1181 |
+
"learning_rate": 6.579358840291064e-07,
|
| 1182 |
+
"logits/chosen": -2.2774598598480225,
|
| 1183 |
+
"logits/rejected": -2.1744918823242188,
|
| 1184 |
+
"logps/chosen": -209.7314453125,
|
| 1185 |
+
"logps/rejected": -224.2617645263672,
|
| 1186 |
+
"loss": 0.5858,
|
| 1187 |
+
"rewards/accuracies": 0.737500011920929,
|
| 1188 |
+
"rewards/chosen": -2.5639281272888184,
|
| 1189 |
+
"rewards/margins": 0.7768001556396484,
|
| 1190 |
+
"rewards/rejected": -3.340728282928467,
|
| 1191 |
+
"step": 800
|
| 1192 |
+
},
|
| 1193 |
+
{
|
| 1194 |
+
"epoch": 0.79,
|
| 1195 |
+
"eval_logits/chosen": -2.159109354019165,
|
| 1196 |
+
"eval_logits/rejected": -2.058640241622925,
|
| 1197 |
+
"eval_logps/chosen": -205.4982452392578,
|
| 1198 |
+
"eval_logps/rejected": -212.2717742919922,
|
| 1199 |
+
"eval_loss": 0.5206710696220398,
|
| 1200 |
+
"eval_rewards/accuracies": 0.7334905862808228,
|
| 1201 |
+
"eval_rewards/chosen": -2.507232666015625,
|
| 1202 |
+
"eval_rewards/margins": 0.9439911842346191,
|
| 1203 |
+
"eval_rewards/rejected": -3.4512243270874023,
|
| 1204 |
+
"eval_runtime": 423.3516,
|
| 1205 |
+
"eval_samples_per_second": 0.987,
|
| 1206 |
+
"eval_steps_per_second": 0.125,
|
| 1207 |
+
"step": 800
|
| 1208 |
+
},
|
| 1209 |
+
{
|
| 1210 |
+
"epoch": 0.8,
|
| 1211 |
+
"learning_rate": 6.00939703563006e-07,
|
| 1212 |
+
"logits/chosen": -2.3462395668029785,
|
| 1213 |
+
"logits/rejected": -2.2609333992004395,
|
| 1214 |
+
"logps/chosen": -191.94070434570312,
|
| 1215 |
+
"logps/rejected": -208.1775665283203,
|
| 1216 |
+
"loss": 0.5793,
|
| 1217 |
+
"rewards/accuracies": 0.699999988079071,
|
| 1218 |
+
"rewards/chosen": -2.5009541511535645,
|
| 1219 |
+
"rewards/margins": 0.8846192359924316,
|
| 1220 |
+
"rewards/rejected": -3.385573625564575,
|
| 1221 |
+
"step": 810
|
| 1222 |
+
},
|
| 1223 |
+
{
|
| 1224 |
+
"epoch": 0.81,
|
| 1225 |
+
"learning_rate": 5.461869057133412e-07,
|
| 1226 |
+
"logits/chosen": -2.386518955230713,
|
| 1227 |
+
"logits/rejected": -2.2898330688476562,
|
| 1228 |
+
"logps/chosen": -198.1256866455078,
|
| 1229 |
+
"logps/rejected": -216.5961456298828,
|
| 1230 |
+
"loss": 0.4895,
|
| 1231 |
+
"rewards/accuracies": 0.699999988079071,
|
| 1232 |
+
"rewards/chosen": -2.2799556255340576,
|
| 1233 |
+
"rewards/margins": 1.0353368520736694,
|
| 1234 |
+
"rewards/rejected": -3.3152928352355957,
|
| 1235 |
+
"step": 820
|
| 1236 |
+
},
|
| 1237 |
+
{
|
| 1238 |
+
"epoch": 0.82,
|
| 1239 |
+
"learning_rate": 4.937421706143497e-07,
|
| 1240 |
+
"logits/chosen": -2.2601094245910645,
|
| 1241 |
+
"logits/rejected": -2.1199710369110107,
|
| 1242 |
+
"logps/chosen": -208.54898071289062,
|
| 1243 |
+
"logps/rejected": -210.0823974609375,
|
| 1244 |
+
"loss": 0.5261,
|
| 1245 |
+
"rewards/accuracies": 0.75,
|
| 1246 |
+
"rewards/chosen": -2.4491305351257324,
|
| 1247 |
+
"rewards/margins": 0.9635556936264038,
|
| 1248 |
+
"rewards/rejected": -3.412686586380005,
|
| 1249 |
+
"step": 830
|
| 1250 |
+
},
|
| 1251 |
+
{
|
| 1252 |
+
"epoch": 0.83,
|
| 1253 |
+
"learning_rate": 4.43667451858166e-07,
|
| 1254 |
+
"logits/chosen": -2.262453079223633,
|
| 1255 |
+
"logits/rejected": -2.0967936515808105,
|
| 1256 |
+
"logps/chosen": -188.5206298828125,
|
| 1257 |
+
"logps/rejected": -202.87240600585938,
|
| 1258 |
+
"loss": 0.4361,
|
| 1259 |
+
"rewards/accuracies": 0.7875000238418579,
|
| 1260 |
+
"rewards/chosen": -2.2963976860046387,
|
| 1261 |
+
"rewards/margins": 1.180870771408081,
|
| 1262 |
+
"rewards/rejected": -3.4772682189941406,
|
| 1263 |
+
"step": 840
|
| 1264 |
+
},
|
| 1265 |
+
{
|
| 1266 |
+
"epoch": 0.84,
|
| 1267 |
+
"learning_rate": 3.9602190330830484e-07,
|
| 1268 |
+
"logits/chosen": -2.2342655658721924,
|
| 1269 |
+
"logits/rejected": -2.130017042160034,
|
| 1270 |
+
"logps/chosen": -200.30343627929688,
|
| 1271 |
+
"logps/rejected": -220.6678466796875,
|
| 1272 |
+
"loss": 0.4885,
|
| 1273 |
+
"rewards/accuracies": 0.7124999761581421,
|
| 1274 |
+
"rewards/chosen": -2.533165454864502,
|
| 1275 |
+
"rewards/margins": 1.0824253559112549,
|
| 1276 |
+
"rewards/rejected": -3.6155905723571777,
|
| 1277 |
+
"step": 850
|
| 1278 |
+
},
|
| 1279 |
+
{
|
| 1280 |
+
"epoch": 0.85,
|
| 1281 |
+
"learning_rate": 3.5086180922049295e-07,
|
| 1282 |
+
"logits/chosen": -2.3764655590057373,
|
| 1283 |
+
"logits/rejected": -2.235109329223633,
|
| 1284 |
+
"logps/chosen": -224.97146606445312,
|
| 1285 |
+
"logps/rejected": -221.2453155517578,
|
| 1286 |
+
"loss": 0.4675,
|
| 1287 |
+
"rewards/accuracies": 0.762499988079071,
|
| 1288 |
+
"rewards/chosen": -2.4995951652526855,
|
| 1289 |
+
"rewards/margins": 1.0894619226455688,
|
| 1290 |
+
"rewards/rejected": -3.589057445526123,
|
| 1291 |
+
"step": 860
|
| 1292 |
+
},
|
| 1293 |
+
{
|
| 1294 |
+
"epoch": 0.86,
|
| 1295 |
+
"learning_rate": 3.0824051775340895e-07,
|
| 1296 |
+
"logits/chosen": -2.380509853363037,
|
| 1297 |
+
"logits/rejected": -2.275191068649292,
|
| 1298 |
+
"logps/chosen": -180.95895385742188,
|
| 1299 |
+
"logps/rejected": -195.82479858398438,
|
| 1300 |
+
"loss": 0.547,
|
| 1301 |
+
"rewards/accuracies": 0.6875,
|
| 1302 |
+
"rewards/chosen": -2.3850598335266113,
|
| 1303 |
+
"rewards/margins": 0.8363308906555176,
|
| 1304 |
+
"rewards/rejected": -3.22139048576355,
|
| 1305 |
+
"step": 870
|
| 1306 |
+
},
|
| 1307 |
+
{
|
| 1308 |
+
"epoch": 0.87,
|
| 1309 |
+
"learning_rate": 2.6820837794786336e-07,
|
| 1310 |
+
"logits/chosen": -2.2321319580078125,
|
| 1311 |
+
"logits/rejected": -2.1590046882629395,
|
| 1312 |
+
"logps/chosen": -204.28173828125,
|
| 1313 |
+
"logps/rejected": -214.01327514648438,
|
| 1314 |
+
"loss": 0.6701,
|
| 1315 |
+
"rewards/accuracies": 0.5249999761581421,
|
| 1316 |
+
"rewards/chosen": -2.5836868286132812,
|
| 1317 |
+
"rewards/margins": 0.4722086787223816,
|
| 1318 |
+
"rewards/rejected": -3.0558953285217285,
|
| 1319 |
+
"step": 880
|
| 1320 |
+
},
|
| 1321 |
+
{
|
| 1322 |
+
"epoch": 0.88,
|
| 1323 |
+
"learning_rate": 2.3081268024887694e-07,
|
| 1324 |
+
"logits/chosen": -2.222381114959717,
|
| 1325 |
+
"logits/rejected": -2.051706075668335,
|
| 1326 |
+
"logps/chosen": -199.63970947265625,
|
| 1327 |
+
"logps/rejected": -208.906005859375,
|
| 1328 |
+
"loss": 0.4139,
|
| 1329 |
+
"rewards/accuracies": 0.8125,
|
| 1330 |
+
"rewards/chosen": -2.3457841873168945,
|
| 1331 |
+
"rewards/margins": 1.2535767555236816,
|
| 1332 |
+
"rewards/rejected": -3.599360942840576,
|
| 1333 |
+
"step": 890
|
| 1334 |
+
},
|
| 1335 |
+
{
|
| 1336 |
+
"epoch": 0.89,
|
| 1337 |
+
"learning_rate": 1.9609760064091044e-07,
|
| 1338 |
+
"logits/chosen": -2.3242409229278564,
|
| 1339 |
+
"logits/rejected": -2.2560477256774902,
|
| 1340 |
+
"logps/chosen": -205.27554321289062,
|
| 1341 |
+
"logps/rejected": -201.7273712158203,
|
| 1342 |
+
"loss": 0.5173,
|
| 1343 |
+
"rewards/accuracies": 0.7124999761581421,
|
| 1344 |
+
"rewards/chosen": -2.428300142288208,
|
| 1345 |
+
"rewards/margins": 0.8561038970947266,
|
| 1346 |
+
"rewards/rejected": -3.2844040393829346,
|
| 1347 |
+
"step": 900
|
| 1348 |
+
},
|
| 1349 |
+
{
|
| 1350 |
+
"epoch": 0.9,
|
| 1351 |
+
"learning_rate": 1.6410414846224992e-07,
|
| 1352 |
+
"logits/chosen": -2.220360517501831,
|
| 1353 |
+
"logits/rejected": -2.109575033187866,
|
| 1354 |
+
"logps/chosen": -201.75045776367188,
|
| 1355 |
+
"logps/rejected": -214.3638458251953,
|
| 1356 |
+
"loss": 0.4565,
|
| 1357 |
+
"rewards/accuracies": 0.7875000238418579,
|
| 1358 |
+
"rewards/chosen": -2.569467067718506,
|
| 1359 |
+
"rewards/margins": 1.1193673610687256,
|
| 1360 |
+
"rewards/rejected": -3.6888339519500732,
|
| 1361 |
+
"step": 910
|
| 1362 |
+
},
|
| 1363 |
+
{
|
| 1364 |
+
"epoch": 0.91,
|
| 1365 |
+
"learning_rate": 1.348701179601819e-07,
|
| 1366 |
+
"logits/chosen": -2.401984691619873,
|
| 1367 |
+
"logits/rejected": -2.2650654315948486,
|
| 1368 |
+
"logps/chosen": -215.353515625,
|
| 1369 |
+
"logps/rejected": -225.1014404296875,
|
| 1370 |
+
"loss": 0.4529,
|
| 1371 |
+
"rewards/accuracies": 0.7875000238418579,
|
| 1372 |
+
"rewards/chosen": -2.38716983795166,
|
| 1373 |
+
"rewards/margins": 1.1682064533233643,
|
| 1374 |
+
"rewards/rejected": -3.5553765296936035,
|
| 1375 |
+
"step": 920
|
| 1376 |
+
},
|
| 1377 |
+
{
|
| 1378 |
+
"epoch": 0.91,
|
| 1379 |
+
"learning_rate": 1.0843004364420151e-07,
|
| 1380 |
+
"logits/chosen": -2.2123489379882812,
|
| 1381 |
+
"logits/rejected": -2.100048065185547,
|
| 1382 |
+
"logps/chosen": -215.25479125976562,
|
| 1383 |
+
"logps/rejected": -229.445556640625,
|
| 1384 |
+
"loss": 0.5697,
|
| 1385 |
+
"rewards/accuracies": 0.6875,
|
| 1386 |
+
"rewards/chosen": -2.393723964691162,
|
| 1387 |
+
"rewards/margins": 0.8360812067985535,
|
| 1388 |
+
"rewards/rejected": -3.2298049926757812,
|
| 1389 |
+
"step": 930
|
| 1390 |
+
},
|
| 1391 |
+
{
|
| 1392 |
+
"epoch": 0.92,
|
| 1393 |
+
"learning_rate": 8.481515948997931e-08,
|
| 1394 |
+
"logits/chosen": -2.3680367469787598,
|
| 1395 |
+
"logits/rejected": -2.26953387260437,
|
| 1396 |
+
"logps/chosen": -216.99514770507812,
|
| 1397 |
+
"logps/rejected": -212.17263793945312,
|
| 1398 |
+
"loss": 0.6119,
|
| 1399 |
+
"rewards/accuracies": 0.737500011920929,
|
| 1400 |
+
"rewards/chosen": -2.478694200515747,
|
| 1401 |
+
"rewards/margins": 0.7769169807434082,
|
| 1402 |
+
"rewards/rejected": -3.255610942840576,
|
| 1403 |
+
"step": 940
|
| 1404 |
+
},
|
| 1405 |
+
{
|
| 1406 |
+
"epoch": 0.93,
|
| 1407 |
+
"learning_rate": 6.4053362042297e-08,
|
| 1408 |
+
"logits/chosen": -2.2625975608825684,
|
| 1409 |
+
"logits/rejected": -2.104025363922119,
|
| 1410 |
+
"logps/chosen": -205.9175262451172,
|
| 1411 |
+
"logps/rejected": -217.52304077148438,
|
| 1412 |
+
"loss": 0.4583,
|
| 1413 |
+
"rewards/accuracies": 0.7749999761581421,
|
| 1414 |
+
"rewards/chosen": -2.4613893032073975,
|
| 1415 |
+
"rewards/margins": 1.17227303981781,
|
| 1416 |
+
"rewards/rejected": -3.633662462234497,
|
| 1417 |
+
"step": 950
|
| 1418 |
+
},
|
| 1419 |
+
{
|
| 1420 |
+
"epoch": 0.94,
|
| 1421 |
+
"learning_rate": 4.616917746052163e-08,
|
| 1422 |
+
"logits/chosen": -2.3516554832458496,
|
| 1423 |
+
"logits/rejected": -2.214130401611328,
|
| 1424 |
+
"logps/chosen": -203.63438415527344,
|
| 1425 |
+
"logps/rejected": -211.27334594726562,
|
| 1426 |
+
"loss": 0.546,
|
| 1427 |
+
"rewards/accuracies": 0.75,
|
| 1428 |
+
"rewards/chosen": -2.5478570461273193,
|
| 1429 |
+
"rewards/margins": 0.990521252155304,
|
| 1430 |
+
"rewards/rejected": -3.5383784770965576,
|
| 1431 |
+
"step": 960
|
| 1432 |
+
},
|
| 1433 |
+
{
|
| 1434 |
+
"epoch": 0.95,
|
| 1435 |
+
"learning_rate": 3.118373254556412e-08,
|
| 1436 |
+
"logits/chosen": -2.3878164291381836,
|
| 1437 |
+
"logits/rejected": -2.2502613067626953,
|
| 1438 |
+
"logps/chosen": -198.12130737304688,
|
| 1439 |
+
"logps/rejected": -199.36111450195312,
|
| 1440 |
+
"loss": 0.4535,
|
| 1441 |
+
"rewards/accuracies": 0.7749999761581421,
|
| 1442 |
+
"rewards/chosen": -2.448812484741211,
|
| 1443 |
+
"rewards/margins": 1.0348026752471924,
|
| 1444 |
+
"rewards/rejected": -3.483614683151245,
|
| 1445 |
+
"step": 970
|
| 1446 |
+
},
|
| 1447 |
+
{
|
| 1448 |
+
"epoch": 0.96,
|
| 1449 |
+
"learning_rate": 1.9114729782535037e-08,
|
| 1450 |
+
"logits/chosen": -2.4108872413635254,
|
| 1451 |
+
"logits/rejected": -2.297142505645752,
|
| 1452 |
+
"logps/chosen": -194.57632446289062,
|
| 1453 |
+
"logps/rejected": -202.7489776611328,
|
| 1454 |
+
"loss": 0.4981,
|
| 1455 |
+
"rewards/accuracies": 0.699999988079071,
|
| 1456 |
+
"rewards/chosen": -2.350036859512329,
|
| 1457 |
+
"rewards/margins": 0.9002410173416138,
|
| 1458 |
+
"rewards/rejected": -3.2502777576446533,
|
| 1459 |
+
"step": 980
|
| 1460 |
+
},
|
| 1461 |
+
{
|
| 1462 |
+
"epoch": 0.97,
|
| 1463 |
+
"learning_rate": 9.97642642858815e-09,
|
| 1464 |
+
"logits/chosen": -2.28024959564209,
|
| 1465 |
+
"logits/rejected": -2.139801502227783,
|
| 1466 |
+
"logps/chosen": -211.3317413330078,
|
| 1467 |
+
"logps/rejected": -211.1747283935547,
|
| 1468 |
+
"loss": 0.5021,
|
| 1469 |
+
"rewards/accuracies": 0.6875,
|
| 1470 |
+
"rewards/chosen": -2.55255126953125,
|
| 1471 |
+
"rewards/margins": 0.9248201251029968,
|
| 1472 |
+
"rewards/rejected": -3.4773712158203125,
|
| 1473 |
+
"step": 990
|
| 1474 |
+
},
|
| 1475 |
+
{
|
| 1476 |
+
"epoch": 0.98,
|
| 1477 |
+
"learning_rate": 3.779617670651436e-09,
|
| 1478 |
+
"logits/chosen": -2.248671054840088,
|
| 1479 |
+
"logits/rejected": -2.1480166912078857,
|
| 1480 |
+
"logps/chosen": -217.71707153320312,
|
| 1481 |
+
"logps/rejected": -225.20040893554688,
|
| 1482 |
+
"loss": 0.6128,
|
| 1483 |
+
"rewards/accuracies": 0.699999988079071,
|
| 1484 |
+
"rewards/chosen": -2.73987078666687,
|
| 1485 |
+
"rewards/margins": 0.9156008958816528,
|
| 1486 |
+
"rewards/rejected": -3.6554713249206543,
|
| 1487 |
+
"step": 1000
|
| 1488 |
+
},
|
| 1489 |
+
{
|
| 1490 |
+
"epoch": 0.98,
|
| 1491 |
+
"eval_logits/chosen": -2.1620335578918457,
|
| 1492 |
+
"eval_logits/rejected": -2.062356948852539,
|
| 1493 |
+
"eval_logps/chosen": -206.584716796875,
|
| 1494 |
+
"eval_logps/rejected": -214.93492126464844,
|
| 1495 |
+
"eval_loss": 0.521207869052887,
|
| 1496 |
+
"eval_rewards/accuracies": 0.7405660152435303,
|
| 1497 |
+
"eval_rewards/chosen": -2.5398268699645996,
|
| 1498 |
+
"eval_rewards/margins": 0.9912916421890259,
|
| 1499 |
+
"eval_rewards/rejected": -3.531118631362915,
|
| 1500 |
+
"eval_runtime": 423.4333,
|
| 1501 |
+
"eval_samples_per_second": 0.987,
|
| 1502 |
+
"eval_steps_per_second": 0.125,
|
| 1503 |
+
"step": 1000
|
| 1504 |
+
},
|
| 1505 |
+
{
|
| 1506 |
+
"epoch": 0.99,
|
| 1507 |
+
"learning_rate": 5.316238729444201e-10,
|
| 1508 |
+
"logits/chosen": -2.234221935272217,
|
| 1509 |
+
"logits/rejected": -2.0627448558807373,
|
| 1510 |
+
"logps/chosen": -209.44503784179688,
|
| 1511 |
+
"logps/rejected": -219.24606323242188,
|
| 1512 |
+
"loss": 0.4684,
|
| 1513 |
+
"rewards/accuracies": 0.7749999761581421,
|
| 1514 |
+
"rewards/chosen": -2.601386547088623,
|
| 1515 |
+
"rewards/margins": 1.190915584564209,
|
| 1516 |
+
"rewards/rejected": -3.792301893234253,
|
| 1517 |
+
"step": 1010
|
| 1518 |
+
},
|
| 1519 |
+
{
|
| 1520 |
+
"epoch": 1.0,
|
| 1521 |
+
"step": 1016,
|
| 1522 |
+
"total_flos": 0.0,
|
| 1523 |
+
"train_loss": 0.5583291621658746,
|
| 1524 |
+
"train_runtime": 16785.8838,
|
| 1525 |
+
"train_samples_per_second": 0.484,
|
| 1526 |
+
"train_steps_per_second": 0.061
|
| 1527 |
+
}
|
| 1528 |
+
],
|
| 1529 |
+
"logging_steps": 10,
|
| 1530 |
+
"max_steps": 1016,
|
| 1531 |
+
"num_input_tokens_seen": 0,
|
| 1532 |
+
"num_train_epochs": 1,
|
| 1533 |
+
"save_steps": 100,
|
| 1534 |
+
"total_flos": 0.0,
|
| 1535 |
+
"train_batch_size": 4,
|
| 1536 |
+
"trial_name": null,
|
| 1537 |
+
"trial_params": null
|
| 1538 |
+
}
|