Hugging Face
Models
Datasets
Spaces
Posts
Docs
Solutions
Pricing
Log In
Sign Up
DUAL-GPO
/
phi-2-gpo-iter-0
like
0
Follow
DUAL Group
2
PEFT
Safetensors
HuggingFaceH4/ultrafeedback_binarized
phi
alignment-handbook
Generated from Trainer
trl
dpo
custom_code
License:
mit
Model card
Files
Files and versions
Community
Use this model
b3fdd14
phi-2-gpo-iter-0
/
all_results.json
lole25
Model save
68b05e6
verified
8 months ago
raw
Copy download link
history
blame
195 Bytes
{
"epoch"
:
1.33
,
"train_loss"
:
0.003023708879482001
,
"train_runtime"
:
17.7261
,
"train_samples"
:
30567
,
"train_samples_per_second"
:
1.128
,
"train_steps_per_second"
:
0.113
}