Hugging Face's logo

mechark
/
outputs

PEFT
Safetensors
trl
sft
unsloth
Generated from Trainer
Model card Files
xet
Community
mechark commited on
Commit
472031e
·
verified ·
1 Parent(s): ad78281

mechark/LLama-8b-Instruct-humanizer

Browse files
Files changed (4) hide show
  1. README.md +36 -16
  2. adapter_config.json +4 -4
  3. adapter_model.safetensors +1 -1
  4. training_args.bin +1 -1
README.md CHANGED
@@ -19,7 +19,7 @@ should probably proofread and complete it, then remove this comment. -->
19
 
20
  This model is a fine-tuned version of [unsloth/llama-3.1-8b-instruct-unsloth-bnb-4bit](https://huggingface.co/unsloth/llama-3.1-8b-instruct-unsloth-bnb-4bit) on an unknown dataset.
21
  It achieves the following results on the evaluation set:
22
- - Loss: 1.6516
23
 
24
  ## Model description
25
 
@@ -38,35 +38,55 @@ More information needed
38
  ### Training hyperparameters
39
 
40
  The following hyperparameters were used during training:
41
- - learning_rate: 1e-05
42
- - train_batch_size: 6
43
  - eval_batch_size: 8
44
  - seed: 3407
45
  - gradient_accumulation_steps: 16
46
- - total_train_batch_size: 96
47
  - optimizer: Use adamw_8bit with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
48
- - lr_scheduler_type: cosine
49
- - lr_scheduler_warmup_steps: 10
50
  - num_epochs: 3
51
 
52
  ### Training results
53
 
54
  | Training Loss | Epoch | Step | Validation Loss |
55
  |:-------------:|:------:|:----:|:---------------:|
56
- | 1.9534 | 0.3042 | 10 | 1.8653 |
57
- | 1.8372 | 0.6084 | 20 | 1.8373 |
58
- | 1.8058 | 0.9125 | 30 | 1.8220 |
59
- | 1.9211 | 1.2433 | 40 | 1.7757 |
60
- | 1.804 | 1.5475 | 50 | 1.7168 |
61
- | 1.7052 | 1.8517 | 60 | 1.6860 |
62
- | 1.6974 | 2.1825 | 70 | 1.6672 |
63
- | 1.7511 | 2.4867 | 80 | 1.6541 |
64
- | 1.6929 | 2.7909 | 90 | 1.6516 |
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
65
 
66
 
67
  ### Framework versions
68
 
69
- - PEFT 0.15.1
70
  - Transformers 4.51.3
71
  - Pytorch 2.6.0+cu124
72
  - Datasets 3.5.0
 
19
 
20
  This model is a fine-tuned version of [unsloth/llama-3.1-8b-instruct-unsloth-bnb-4bit](https://huggingface.co/unsloth/llama-3.1-8b-instruct-unsloth-bnb-4bit) on an unknown dataset.
21
  It achieves the following results on the evaluation set:
22
+ - Loss: 1.3917
23
 
24
  ## Model description
25
 
 
38
  ### Training hyperparameters
39
 
40
  The following hyperparameters were used during training:
41
+ - learning_rate: 2e-05
42
+ - train_batch_size: 2
43
  - eval_batch_size: 8
44
  - seed: 3407
45
  - gradient_accumulation_steps: 16
46
+ - total_train_batch_size: 32
47
  - optimizer: Use adamw_8bit with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
48
+ - lr_scheduler_type: linear
49
+ - lr_scheduler_warmup_ratio: 0.05
50
  - num_epochs: 3
51
 
52
  ### Training results
53
 
54
  | Training Loss | Epoch | Step | Validation Loss |
55
  |:-------------:|:------:|:----:|:---------------:|
56
+ | 1.8249 | 0.1014 | 10 | 1.7483 |
57
+ | 1.7082 | 0.2028 | 20 | 1.6279 |
58
+ | 1.6314 | 0.3042 | 30 | 1.5369 |
59
+ | 1.5673 | 0.4056 | 40 | 1.4923 |
60
+ | 1.5182 | 0.5070 | 50 | 1.4698 |
61
+ | 1.4977 | 0.6084 | 60 | 1.4576 |
62
+ | 1.4762 | 0.7098 | 70 | 1.4489 |
63
+ | 1.4641 | 0.8112 | 80 | 1.4424 |
64
+ | 1.4763 | 0.9125 | 90 | 1.4365 |
65
+ | 1.5972 | 1.0203 | 100 | 1.4317 |
66
+ | 1.4753 | 1.1217 | 110 | 1.4267 |
67
+ | 1.4513 | 1.2231 | 120 | 1.4224 |
68
+ | 1.4775 | 1.3245 | 130 | 1.4188 |
69
+ | 1.4559 | 1.4259 | 140 | 1.4156 |
70
+ | 1.4214 | 1.5272 | 150 | 1.4126 |
71
+ | 1.4534 | 1.6286 | 160 | 1.4098 |
72
+ | 1.4176 | 1.7300 | 170 | 1.4073 |
73
+ | 1.4204 | 1.8314 | 180 | 1.4046 |
74
+ | 1.4398 | 1.9328 | 190 | 1.4027 |
75
+ | 1.5789 | 2.0406 | 200 | 1.4010 |
76
+ | 1.4298 | 2.1420 | 210 | 1.3990 |
77
+ | 1.4299 | 2.2433 | 220 | 1.3974 |
78
+ | 1.4136 | 2.3447 | 230 | 1.3959 |
79
+ | 1.4535 | 2.4461 | 240 | 1.3949 |
80
+ | 1.4244 | 2.5475 | 250 | 1.3938 |
81
+ | 1.4079 | 2.6489 | 260 | 1.3929 |
82
+ | 1.4013 | 2.7503 | 270 | 1.3922 |
83
+ | 1.4204 | 2.8517 | 280 | 1.3918 |
84
+ | 1.4224 | 2.9531 | 290 | 1.3917 |
85
 
86
 
87
  ### Framework versions
88
 
89
+ - PEFT 0.15.2
90
  - Transformers 4.51.3
91
  - Pytorch 2.6.0+cu124
92
  - Datasets 3.5.0
adapter_config.json CHANGED
@@ -24,13 +24,13 @@
24
  "rank_pattern": {},
25
  "revision": null,
26
  "target_modules": [
27
- "o_proj",
28
- "down_proj",
29
- "q_proj",
30
  "v_proj",
31
  "k_proj",
 
 
 
32
  "up_proj",
33
- "gate_proj"
34
  ],
35
  "task_type": "CAUSAL_LM",
36
  "trainable_token_indices": null,
 
24
  "rank_pattern": {},
25
  "revision": null,
26
  "target_modules": [
 
 
 
27
  "v_proj",
28
  "k_proj",
29
+ "down_proj",
30
+ "gate_proj",
31
+ "o_proj",
32
  "up_proj",
33
+ "q_proj"
34
  ],
35
  "task_type": "CAUSAL_LM",
36
  "trainable_token_indices": null,
adapter_model.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:aae4b0d50da6e41e40ba2d8feb7859f753cb4bb36ac239019c00adaade210a8a
3
  size 167832240
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:18b1fbfec421ee0b121ddb4e44942ef050646f3037e10fabd4a109bc9379f0a1
3
  size 167832240
training_args.bin CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:e41b177df17c65dcf90fa1340952e9e13ce4209db9d0a643a6b1b642df3f5efe
3
  size 5304
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:65f812ab14de6705ab7c2fd93b407c5256019da270f903ab42c79889bdbf1255
3
  size 5304