Hugging Face's logo

sizhkhy
/
centime

PEFT
Safetensors
llama-factory
lora
unsloth
Generated from Trainer
Model card Files Community
sizhkhy commited on
Commit
26cdf89
·
verified ·
1 Parent(s): 2147dcc

1.5

Browse files
Files changed (10) hide show
  1. README.md +43 -4
  2. adapter_config.json +4 -4
  3. adapter_model.safetensors +1 -1
  4. all_results.json +10 -10
  5. eval_results.json +5 -5
  6. train_results.json +6 -6
  7. trainer_log.jsonl +0 -0
  8. trainer_state.json +0 -0
  9. training_args.bin +1 -1
  10. training_loss.png +0 -0
README.md CHANGED
@@ -8,18 +8,18 @@ tags:
8
  - unsloth
9
  - generated_from_trainer
10
  model-index:
11
- - name: llm3br256-v1.6.dummy
12
  results: []
13
  ---
14
 
15
  <!-- This model card has been generated automatically according to the information the Trainer had access to. You
16
  should probably proofread and complete it, then remove this comment. -->
17
 
18
- # llm3br256-v1.6.dummy
19
 
20
  This model is a fine-tuned version of [meta-llama/Llama-3.2-3B-Instruct](https://huggingface.co/meta-llama/Llama-3.2-3B-Instruct) on the centime dataset.
21
  It achieves the following results on the evaluation set:
22
- - Loss: 0.0866
23
 
24
  ## Model description
25
 
@@ -45,10 +45,49 @@ The following hyperparameters were used during training:
45
  - optimizer: Use OptimizerNames.ADAMW_TORCH with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
46
  - lr_scheduler_type: cosine
47
  - lr_scheduler_warmup_ratio: 0.1
48
- - num_epochs: 1
49
 
50
  ### Training results
51
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
52
 
53
 
54
  ### Framework versions
 
8
  - unsloth
9
  - generated_from_trainer
10
  model-index:
11
+ - name: llm3br256
12
  results: []
13
  ---
14
 
15
  <!-- This model card has been generated automatically according to the information the Trainer had access to. You
16
  should probably proofread and complete it, then remove this comment. -->
17
 
18
+ # llm3br256
19
 
20
  This model is a fine-tuned version of [meta-llama/Llama-3.2-3B-Instruct](https://huggingface.co/meta-llama/Llama-3.2-3B-Instruct) on the centime dataset.
21
  It achieves the following results on the evaluation set:
22
+ - Loss: 0.0070
23
 
24
  ## Model description
25
 
 
45
  - optimizer: Use OptimizerNames.ADAMW_TORCH with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
46
  - lr_scheduler_type: cosine
47
  - lr_scheduler_warmup_ratio: 0.1
48
+ - num_epochs: 25.0
49
 
50
  ### Training results
51
 
52
+ | Training Loss | Epoch | Step | Validation Loss |
53
+ |:-------------:|:------:|:----:|:---------------:|
54
+ | 0.1159 | 0.1208 | 25 | 0.1004 |
55
+ | 0.0843 | 0.2415 | 50 | 0.0635 |
56
+ | 0.0763 | 0.3623 | 75 | 0.0474 |
57
+ | 0.0496 | 0.4831 | 100 | 0.0365 |
58
+ | 0.046 | 0.6039 | 125 | 0.0316 |
59
+ | 0.0368 | 0.7246 | 150 | 0.0266 |
60
+ | 0.0283 | 0.8454 | 175 | 0.0232 |
61
+ | 0.0237 | 0.9662 | 200 | 0.0212 |
62
+ | 0.0234 | 1.0870 | 225 | 0.0194 |
63
+ | 0.0232 | 1.2077 | 250 | 0.0176 |
64
+ | 0.0307 | 1.3285 | 275 | 0.0178 |
65
+ | 0.0228 | 1.4493 | 300 | 0.0147 |
66
+ | 0.0167 | 1.5700 | 325 | 0.0155 |
67
+ | 0.0238 | 1.6908 | 350 | 0.0125 |
68
+ | 0.0191 | 1.8116 | 375 | 0.0138 |
69
+ | 0.0273 | 1.9324 | 400 | 0.0120 |
70
+ | 0.0194 | 2.0531 | 425 | 0.0125 |
71
+ | 0.0125 | 2.1739 | 450 | 0.0128 |
72
+ | 0.0132 | 2.2947 | 475 | 0.0117 |
73
+ | 0.0142 | 2.4155 | 500 | 0.0099 |
74
+ | 0.0119 | 2.5362 | 525 | 0.0105 |
75
+ | 0.0131 | 2.6570 | 550 | 0.0118 |
76
+ | 0.0089 | 2.7778 | 575 | 0.0100 |
77
+ | 0.0158 | 2.8986 | 600 | 0.0096 |
78
+ | 0.0119 | 3.0193 | 625 | 0.0096 |
79
+ | 0.0097 | 3.1401 | 650 | 0.0099 |
80
+ | 0.0089 | 3.2609 | 675 | 0.0092 |
81
+ | 0.0087 | 3.3816 | 700 | 0.0088 |
82
+ | 0.0083 | 3.5024 | 725 | 0.0088 |
83
+ | 0.0088 | 3.6232 | 750 | 0.0080 |
84
+ | 0.0058 | 3.7440 | 775 | 0.0069 |
85
+ | 0.008 | 3.8647 | 800 | 0.0070 |
86
+ | 0.0099 | 3.9855 | 825 | 0.0073 |
87
+ | 0.0072 | 4.1063 | 850 | 0.0113 |
88
+ | 0.0065 | 4.2271 | 875 | 0.0107 |
89
+ | 0.0079 | 4.3478 | 900 | 0.0097 |
90
+ | 0.0081 | 4.4686 | 925 | 0.0103 |
91
 
92
 
93
  ### Framework versions
adapter_config.json CHANGED
@@ -20,13 +20,13 @@
20
  "rank_pattern": {},
21
  "revision": null,
22
  "target_modules": [
23
- "gate_proj",
24
  "down_proj",
25
- "k_proj",
26
- "q_proj",
27
  "o_proj",
28
  "up_proj",
29
- "v_proj"
 
 
 
30
  ],
31
  "task_type": "CAUSAL_LM",
32
  "use_dora": false,
 
20
  "rank_pattern": {},
21
  "revision": null,
22
  "target_modules": [
 
23
  "down_proj",
 
 
24
  "o_proj",
25
  "up_proj",
26
+ "k_proj",
27
+ "v_proj",
28
+ "q_proj",
29
+ "gate_proj"
30
  ],
31
  "task_type": "CAUSAL_LM",
32
  "use_dora": false,
adapter_model.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:0249b678235fe5861e05868132a466f42efcbdf9df92274529610922f15002d4
3
  size 1556140392
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:f546425f4fea154c38e487f0a187535ac63b41c20c7f3dbe76a4f17048ae2e89
3
  size 1556140392
all_results.json CHANGED
@@ -1,12 +1,12 @@
1
  {
2
- "epoch": 1.0,
3
- "eval_loss": 0.0865735113620758,
4
- "eval_runtime": 1.7555,
5
- "eval_samples_per_second": 1.709,
6
- "eval_steps_per_second": 0.57,
7
- "total_flos": 467769719488512.0,
8
- "train_loss": 0.2238682359457016,
9
- "train_runtime": 24.0578,
10
- "train_samples_per_second": 0.291,
11
- "train_steps_per_second": 0.042
12
  }
 
1
  {
2
+ "epoch": 4.468599033816425,
3
+ "eval_loss": 0.007014152128249407,
4
+ "eval_runtime": 20.2851,
5
+ "eval_samples_per_second": 4.93,
6
+ "eval_steps_per_second": 0.148,
7
+ "total_flos": 3.488842745919701e+18,
8
+ "train_loss": 0.025700105609926017,
9
+ "train_runtime": 24639.0069,
10
+ "train_samples_per_second": 10.045,
11
+ "train_steps_per_second": 0.21
12
  }
eval_results.json CHANGED
@@ -1,7 +1,7 @@
1
  {
2
- "epoch": 1.0,
3
- "eval_loss": 0.0865735113620758,
4
- "eval_runtime": 1.7555,
5
- "eval_samples_per_second": 1.709,
6
- "eval_steps_per_second": 0.57
7
  }
 
1
  {
2
+ "epoch": 4.468599033816425,
3
+ "eval_loss": 0.007014152128249407,
4
+ "eval_runtime": 20.2851,
5
+ "eval_samples_per_second": 4.93,
6
+ "eval_steps_per_second": 0.148
7
  }
train_results.json CHANGED
@@ -1,8 +1,8 @@
1
  {
2
- "epoch": 1.0,
3
- "total_flos": 467769719488512.0,
4
- "train_loss": 0.2238682359457016,
5
- "train_runtime": 24.0578,
6
- "train_samples_per_second": 0.291,
7
- "train_steps_per_second": 0.042
8
  }
 
1
  {
2
+ "epoch": 4.468599033816425,
3
+ "total_flos": 3.488842745919701e+18,
4
+ "train_loss": 0.025700105609926017,
5
+ "train_runtime": 24639.0069,
6
+ "train_samples_per_second": 10.045,
7
+ "train_steps_per_second": 0.21
8
  }
trainer_log.jsonl CHANGED
The diff for this file is too large to render. See raw diff
 
trainer_state.json CHANGED
The diff for this file is too large to render. See raw diff
 
training_args.bin CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:266b9d034000aad94dddfd7f25ce54cf844e9fe81339d42693ba718c6736bc9a
3
  size 5432
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:4f09096e9b8b6a21e07e8d8e7f67e8454fecdc716a04339865100ade36fb3fcd
3
  size 5432
training_loss.png CHANGED