Hugging Face's logo

DUAL-GPO
/
phi-2-gpo-20k-40k-60k-lora512-i2

PEFT
TensorBoard
Safetensors
phi
alignment-handbook
Generated from Trainer
trl
dpo
custom_code
Model card Files Metrics Community
BraylonDash commited on
Commit
f61ddb0
•
1 Parent(s): 3780c41

Model save

Browse files
Files changed (7) hide show
  1. README.md +2 -5
  2. adapter_config.json +2 -2
  3. all_results.json +4 -4
  4. runs/May18_18-44-20_ip-10-192-15-31/events.out.tfevents.1716058048.ip-10-192-15-31.2560.0 +3 -0
  5. train_results.json +4 -4
  6. trainer_state.json +4 -4
  7. training_args.bin +1 -1
README.md CHANGED
@@ -1,13 +1,10 @@
1
  ---
2
  library_name: peft
3
  tags:
4
- - alignment-handbook
5
- - generated_from_trainer
6
  - trl
7
  - dpo
 
8
  base_model: DUAL-GPO-2/phi-2-gpo-v34-merged-i1
9
- datasets:
10
- - HuggingFaceH4/ultrafeedback_binarized
11
  model-index:
12
  - name: phi-2-gpo-20k-40k-60k-lora512-i2
13
  results: []
@@ -18,7 +15,7 @@ should probably proofread and complete it, then remove this comment. -->
18
 
19
  # phi-2-gpo-20k-40k-60k-lora512-i2
20
 
21
- This model is a fine-tuned version of [DUAL-GPO-2/phi-2-gpo-v34-merged-i1](https://huggingface.co/DUAL-GPO-2/phi-2-gpo-v34-merged-i1) on the HuggingFaceH4/ultrafeedback_binarized dataset.
22
 
23
  ## Model description
24
 
 
1
  ---
2
  library_name: peft
3
  tags:
 
 
4
  - trl
5
  - dpo
6
+ - generated_from_trainer
7
  base_model: DUAL-GPO-2/phi-2-gpo-v34-merged-i1
 
 
8
  model-index:
9
  - name: phi-2-gpo-20k-40k-60k-lora512-i2
10
  results: []
 
15
 
16
  # phi-2-gpo-20k-40k-60k-lora512-i2
17
 
18
+ This model is a fine-tuned version of [DUAL-GPO-2/phi-2-gpo-v34-merged-i1](https://huggingface.co/DUAL-GPO-2/phi-2-gpo-v34-merged-i1) on the None dataset.
19
 
20
  ## Model description
21
 
adapter_config.json CHANGED
@@ -21,8 +21,8 @@
21
  "target_modules": [
22
  "q_proj",
23
  "k_proj",
24
- "v_proj",
25
- "dense"
26
  ],
27
  "task_type": "CAUSAL_LM"
28
  }
 
21
  "target_modules": [
22
  "q_proj",
23
  "k_proj",
24
+ "dense",
25
+ "v_proj"
26
  ],
27
  "task_type": "CAUSAL_LM"
28
  }
all_results.json CHANGED
@@ -1,8 +1,8 @@
1
  {
2
  "epoch": 1.0,
3
- "train_loss": 0.12654017809873971,
4
- "train_runtime": 11344.5242,
5
  "train_samples": 61135,
6
- "train_samples_per_second": 1.763,
7
- "train_steps_per_second": 0.028
8
  }
 
1
  {
2
  "epoch": 1.0,
3
+ "train_loss": 0.004328498234733557,
4
+ "train_runtime": 430.9957,
5
  "train_samples": 61135,
6
+ "train_samples_per_second": 46.404,
7
+ "train_steps_per_second": 0.724
8
  }
runs/May18_18-44-20_ip-10-192-15-31/events.out.tfevents.1716058048.ip-10-192-15-31.2560.0 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:6736606b52b20cc9f7741fe308e0b821bce3410df12b4d881c9a03d690940225
3
+ size 5893
train_results.json CHANGED
@@ -1,8 +1,8 @@
1
  {
2
  "epoch": 1.0,
3
- "train_loss": 0.12654017809873971,
4
- "train_runtime": 11344.5242,
5
  "train_samples": 61135,
6
- "train_samples_per_second": 1.763,
7
- "train_steps_per_second": 0.028
8
  }
 
1
  {
2
  "epoch": 1.0,
3
+ "train_loss": 0.004328498234733557,
4
+ "train_runtime": 430.9957,
5
  "train_samples": 61135,
6
+ "train_samples_per_second": 46.404,
7
+ "train_steps_per_second": 0.724
8
  }
trainer_state.json CHANGED
@@ -460,10 +460,10 @@
460
  "epoch": 1.0,
461
  "step": 312,
462
  "total_flos": 0.0,
463
- "train_loss": 0.12654017809873971,
464
- "train_runtime": 11344.5242,
465
- "train_samples_per_second": 1.763,
466
- "train_steps_per_second": 0.028
467
  }
468
  ],
469
  "logging_steps": 10,
 
460
  "epoch": 1.0,
461
  "step": 312,
462
  "total_flos": 0.0,
463
+ "train_loss": 0.004328498234733557,
464
+ "train_runtime": 430.9957,
465
+ "train_samples_per_second": 46.404,
466
+ "train_steps_per_second": 0.724
467
  }
468
  ],
469
  "logging_steps": 10,
training_args.bin CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:a40c6e0b07f0202e23d909bf908663a28caeb57e248700e675db08c7dec827ab
3
  size 5880
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:06e9a88bb53184451749fdd4cb47ba77e2d476f723fde7dd5cffd5d1f46aaa6f
3
  size 5880