metadata

base_model: huggingface/CodeBERTa-small-v1
tags:
  - generated_from_trainer
metrics:
  - accuracy
datasets:
  - ejschwartz/oo-method-test-split
widget:
  - text: ' L1: 55 ?? push ebp 8b ec ?? mov ebp, esp 5d ?? pop ebp c3 ?? ret '
    example_title: Function 1
  - text: >2
       L1: 55 ?? push ebp 8b ec ?? mov ebp, esp 51 ?? push ecx a1 b0 5d 43 00 ?? mov eax, dword ds:[0x00435db0] 83 f8 fe ?? cmp eax, 0xfe<254,-2> 75 0a ?? jne basic block L4
      L2: e8 4e 17 00 00 ?? call function 0x00415d25

      L3: a1 b0 5d 43 00 ?? mov eax, dword ds:[0x00435db0]

      L4: 83 f8 ff ?? cmp eax, 0xff<255,-1> 75 07 ?? jne basic block L6

      L5: b8 ff ff 00 00 ?? mov eax, 0x0000ffff eb 1b ?? jmp basic block L9

      L6: 6a 00 ?? push 0 8d 4d fc ?? lea ecx, ss:[ebp + 0xfc<252,-4>] 51 ??
      push ecx 6a 01 ?? push 1 8d 4d 08 ?? lea ecx, ss:[ebp + 8] 51 ?? push ecx
      50 ?? push eax ff 15 28 91 43 00 ?? call dword ds:[0x00439128]

      L7: 85 c0 ?? test eax, eax 74 e2 ?? je basic block L5

      L8: 66 8b 45 08 ?? mov ax, word ss:[ebp + 8]

      L9: 8b e5 ?? mov esp, ebp 5d ?? pop ebp c3 ?? ret
    example_title: Function 2
  - text: >2
       L1: 0f b7 41 32 ?? movzx eax, word ds:[ecx + 0x32<50>] 83 e8 20 ?? sub eax, 0x20<32> 74 2d ?? je basic block L10
      L2: 83 e8 03 ?? sub eax, 3 74 22 ?? je basic block L9

      L3: 83 e8 08 ?? sub eax, 8 74 17 ?? je basic block L8

      L4: 48 ?? dec eax 83 e8 01 ?? sub eax, 1 74 0b ?? je basic block L7

      L5: 83 e8 03 ?? sub eax, 3 75 1c ?? jne basic block L11

      L6: 83 49 20 08 ?? or dword ds:[ecx + 0x20<32>], 8 eb 16 ?? jmp basic
      block L11

      L7: 83 49 20 04 ?? or dword ds:[ecx + 0x20<32>], 4 eb 10 ?? jmp basic
      block L11

      L8: 83 49 20 01 ?? or dword ds:[ecx + 0x20<32>], 1 eb 0a ?? jmp basic
      block L11

      L9: 83 49 20 20 ?? or dword ds:[ecx + 0x20<32>], 0x20<32> eb 04 ?? jmp
      basic block L11

      L10: 83 49 20 02 ?? or dword ds:[ecx + 0x20<32>], 2

      L11: b0 01 ?? mov al, 1 c3 ?? ret
    example_title: Method 1
  - text: >2
       L1: 8b ff ?? mov edi, edi 55 ?? push ebp 8b ec ?? mov ebp, esp 83 ec 08 ?? sub esp, 8 89 4d f8 ?? mov dword ss:[ebp + 0xf8<248,-8>], ecx 8b 4d f8 ?? mov ecx, dword ss:[ebp + 0xf8<248,-8>] e8 e6 ac f9 ff ?? call function 0x00401569
      L2: 23 45 08 ?? and eax, dword ss:[ebp + 8] 3b 45 08 ?? cmp eax, dword
      ss:[ebp + 8] 75 09 ?? jne basic block L4

      L3: c7 45 fc 01 00 00 00 ?? mov dword ss:[ebp + 0xfc<252,-4>], 1 eb 07 ??
      jmp basic block L5

      L4: c7 45 fc 00 00 00 00 ?? mov dword ss:[ebp + 0xfc<252,-4>], 0

      L5: 8a 45 fc ?? mov al, byte ss:[ebp + 0xfc<252,-4>] 8b e5 ?? mov esp, ebp
      5d ?? pop ebp c2 04 00 ?? ret 4
    example_title: Method 2

oo-method-test-model-bylibrary

This model is a fine-tuned version of huggingface/CodeBERTa-small-v1 on the ejschwartz/oo-method-test-split dataset. It achieves the following results on the evaluation set:

Loss: 0.3303
Accuracy: 0.9161
Best Accuracy: 0.9161

Model description

More information needed

Intended uses & limitations

More information needed

Training and evaluation data

More information needed

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

learning_rate: 2.386135927313411e-05
train_batch_size: 8
eval_batch_size: 8
seed: 42
optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
lr_scheduler_type: cosine
lr_scheduler_warmup_ratio: 0.05
training_steps: 887

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	Best Accuracy
0.3788	0.02	178	0.6188	0.8470	0.8470
0.1456	0.04	356	0.6572	0.8519	0.8519
0.17	0.05	534	0.4926	0.8798	0.8798
0.1162	0.07	712	0.3303	0.9161	0.9161

Framework versions

Transformers 4.33.1
Pytorch 2.0.1+cu117
Datasets 2.14.5
Tokenizers 0.13.3