Hugging Face's logo

Basha738
/
outputs

PEFT
TensorBoard
Safetensors
trl
sft
Generated from Trainer
Model card Files Metrics Community
Basha738 commited on
Commit
9daabf3
·
verified ·
1 Parent(s): bb9c6e2

llama2-supervised-ft-5epochs-411

Browse files
Files changed (6) hide show
  1. README.md +24 -24
  2. adapter_config.json +3 -3
  3. adapter_model.safetensors +2 -2
  4. runs/Feb09_03-46-53_7530220781bc/events.out.tfevents.1707450416.7530220781bc.616.0 +3 -0
  5. runs/Feb09_04-24-32_7530220781bc/events.out.tfevents.1707452682.7530220781bc.11420.0 +3 -0
  6. training_args.bin +1 -1
README.md CHANGED
@@ -17,7 +17,7 @@ should probably proofread and complete it, then remove this comment. -->
17
 
18
  This model is a fine-tuned version of [Trelis/Llama-2-7b-chat-hf-sharded-bf16](https://huggingface.co/Trelis/Llama-2-7b-chat-hf-sharded-bf16) on the None dataset.
19
  It achieves the following results on the evaluation set:
20
- - Loss: 2.1986
21
 
22
  ## Model description
23
 
@@ -40,8 +40,8 @@ The following hyperparameters were used during training:
40
  - train_batch_size: 1
41
  - eval_batch_size: 1
42
  - seed: 42
43
- - gradient_accumulation_steps: 4
44
- - total_train_batch_size: 4
45
  - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
46
  - lr_scheduler_type: linear
47
  - num_epochs: 5
@@ -51,27 +51,27 @@ The following hyperparameters were used during training:
51
 
52
  | Training Loss | Epoch | Step | Validation Loss |
53
  |:-------------:|:-----:|:----:|:---------------:|
54
- | 2.591 | 0.23 | 4 | 2.3712 |
55
- | 2.6663 | 0.46 | 8 | 2.3530 |
56
- | 2.4782 | 0.69 | 12 | 2.3351 |
57
- | 2.2831 | 0.91 | 16 | 2.3186 |
58
- | 2.4023 | 1.14 | 20 | 2.3037 |
59
- | 2.2617 | 1.37 | 24 | 2.2902 |
60
- | 2.2799 | 1.6 | 28 | 2.2779 |
61
- | 2.3633 | 1.83 | 32 | 2.2665 |
62
- | 2.2249 | 2.06 | 36 | 2.2560 |
63
- | 2.5186 | 2.29 | 40 | 2.2464 |
64
- | 2.4234 | 2.51 | 44 | 2.2379 |
65
- | 2.26 | 2.74 | 48 | 2.2303 |
66
- | 2.2937 | 2.97 | 52 | 2.2237 |
67
- | 2.3681 | 3.2 | 56 | 2.2179 |
68
- | 2.095 | 3.43 | 60 | 2.2129 |
69
- | 2.1456 | 3.66 | 64 | 2.2087 |
70
- | 2.4307 | 3.89 | 68 | 2.2053 |
71
- | 2.2769 | 4.11 | 72 | 2.2025 |
72
- | 2.3649 | 4.34 | 76 | 2.2005 |
73
- | 2.3577 | 4.57 | 80 | 2.1992 |
74
- | 2.1575 | 4.8 | 84 | 2.1986 |
75
 
76
 
77
  ### Framework versions
 
17
 
18
  This model is a fine-tuned version of [Trelis/Llama-2-7b-chat-hf-sharded-bf16](https://huggingface.co/Trelis/Llama-2-7b-chat-hf-sharded-bf16) on the None dataset.
19
  It achieves the following results on the evaluation set:
20
+ - Loss: 2.1177
21
 
22
  ## Model description
23
 
 
40
  - train_batch_size: 1
41
  - eval_batch_size: 1
42
  - seed: 42
43
+ - gradient_accumulation_steps: 16
44
+ - total_train_batch_size: 16
45
  - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
46
  - lr_scheduler_type: linear
47
  - num_epochs: 5
 
51
 
52
  | Training Loss | Epoch | Step | Validation Loss |
53
  |:-------------:|:-----:|:----:|:---------------:|
54
+ | 2.3322 | 0.24 | 4 | 2.3370 |
55
+ | 2.2294 | 0.47 | 8 | 2.3072 |
56
+ | 2.343 | 0.71 | 12 | 2.2808 |
57
+ | 2.3434 | 0.94 | 16 | 2.2575 |
58
+ | 2.3287 | 1.18 | 20 | 2.2372 |
59
+ | 2.189 | 1.41 | 24 | 2.2194 |
60
+ | 2.2015 | 1.65 | 28 | 2.2040 |
61
+ | 2.1622 | 1.88 | 32 | 2.1907 |
62
+ | 2.1723 | 2.12 | 36 | 2.1790 |
63
+ | 2.1742 | 2.35 | 40 | 2.1687 |
64
+ | 2.0336 | 2.59 | 44 | 2.1597 |
65
+ | 2.1124 | 2.82 | 48 | 2.1518 |
66
+ | 2.1201 | 3.06 | 52 | 2.1449 |
67
+ | 2.1692 | 3.29 | 56 | 2.1387 |
68
+ | 2.2336 | 3.53 | 60 | 2.1334 |
69
+ | 2.0764 | 3.76 | 64 | 2.1288 |
70
+ | 2.1874 | 4.0 | 68 | 2.1250 |
71
+ | 2.1178 | 4.24 | 72 | 2.1220 |
72
+ | 2.1511 | 4.47 | 76 | 2.1199 |
73
+ | 2.1382 | 4.71 | 80 | 2.1184 |
74
+ | 2.1022 | 4.94 | 84 | 2.1177 |
75
 
76
 
77
  ### Framework versions
adapter_config.json CHANGED
@@ -9,21 +9,21 @@
9
  "layers_pattern": null,
10
  "layers_to_transform": null,
11
  "loftq_config": {},
12
- "lora_alpha": 32,
13
  "lora_dropout": 0.05,
14
  "megatron_config": null,
15
  "megatron_core": "megatron.core",
16
  "modules_to_save": null,
17
  "peft_type": "LORA",
18
- "r": 16,
19
  "rank_pattern": {},
20
  "revision": null,
21
  "target_modules": [
22
  "gate_proj",
23
  "o_proj",
24
- "down_proj",
25
  "up_proj",
26
  "q_proj",
 
27
  "k_proj",
28
  "v_proj"
29
  ],
 
9
  "layers_pattern": null,
10
  "layers_to_transform": null,
11
  "loftq_config": {},
12
+ "lora_alpha": 64,
13
  "lora_dropout": 0.05,
14
  "megatron_config": null,
15
  "megatron_core": "megatron.core",
16
  "modules_to_save": null,
17
  "peft_type": "LORA",
18
+ "r": 32,
19
  "rank_pattern": {},
20
  "revision": null,
21
  "target_modules": [
22
  "gate_proj",
23
  "o_proj",
 
24
  "up_proj",
25
  "q_proj",
26
+ "down_proj",
27
  "k_proj",
28
  "v_proj"
29
  ],
adapter_model.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:dfd60a3d7392d23dd321852d451e4f83d01000f6df1f6fb40002ed5071608111
3
- size 159967880
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:45b285036211d3480b33717cc2cd70493d538455f8372fd4d1932dbf82b9a67e
3
+ size 319876032
runs/Feb09_03-46-53_7530220781bc/events.out.tfevents.1707450416.7530220781bc.616.0 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:8158ba849aeca2198f472c3bfa781e046a06de73f9b35e3bbe255647ade4a265
3
+ size 7286
runs/Feb09_04-24-32_7530220781bc/events.out.tfevents.1707452682.7530220781bc.11420.0 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:14771349895a4cf91974475d8e168c01abab9479ffcc1b5b14c238ab8fd2f1c5
3
+ size 23930
training_args.bin CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:e7a347c11818060515daa22efc2808605735815351a1166cecf871f4c53ff334
3
  size 4728
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:b9fee88b62fd20e4d5eb24648a8c3a2fcab0d481839c0fc33b7b99a961ed4cd6
3
  size 4728