Hugging Face's logo

martimfasantos
/
tinyllama-1.1b-chat-dpo-full

Text Generation
Transformers
TensorBoard
Safetensors
llama
alignment-handbook
trl
dpo
Generated from Trainer
conversational
text-generation-inference
Inference Endpoints
Model card Files Metrics Community
martimfasantos commited on
Commit
309f872
•
1 Parent(s): f7851cd

Model save

Browse files
Files changed (7) hide show
  1. README.md +108 -0
  2. all_results.json +9 -0
  3. generation_config.json +7 -0
  4. model.safetensors +1 -1
  5. runs/Jun02_10-42-23_poseidon/events.out.tfevents.1717325399.poseidon.2371184.0 +2 -2
  6. train_results.json +9 -0
  7. trainer_state.json +0 -0
README.md ADDED
@@ -0,0 +1,108 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ license: apache-2.0
3
+ base_model: martimfasantos/tinyllama-1.1b-chat-sft-full
4
+ tags:
5
+ - trl
6
+ - dpo
7
+ - generated_from_trainer
8
+ model-index:
9
+ - name: tinyllama-1.1b-chat-dpo-full
10
+ results: []
11
+ ---
12
+
13
+ <!-- This model card has been generated automatically according to the information the Trainer had access to. You
14
+ should probably proofread and complete it, then remove this comment. -->
15
+
16
+ # tinyllama-1.1b-chat-dpo-full
17
+
18
+ This model is a fine-tuned version of [martimfasantos/tinyllama-1.1b-chat-sft-full](https://huggingface.co/martimfasantos/tinyllama-1.1b-chat-sft-full) on an unknown dataset.
19
+ It achieves the following results on the evaluation set:
20
+ - Loss: 0.5862
21
+ - Rewards/chosen: -1.1608
22
+ - Rewards/rejected: -1.6138
23
+ - Rewards/accuracies: 0.6885
24
+ - Rewards/margins: 0.4530
25
+ - Logps/rejected: -458.4823
26
+ - Logps/chosen: -452.2973
27
+ - Logits/rejected: -2.3882
28
+ - Logits/chosen: -2.4306
29
+
30
+ ## Model description
31
+
32
+ More information needed
33
+
34
+ ## Intended uses & limitations
35
+
36
+ More information needed
37
+
38
+ ## Training and evaluation data
39
+
40
+ More information needed
41
+
42
+ ## Training procedure
43
+
44
+ ### Training hyperparameters
45
+
46
+ The following hyperparameters were used during training:
47
+ - learning_rate: 5e-07
48
+ - train_batch_size: 4
49
+ - eval_batch_size: 8
50
+ - seed: 42
51
+ - distributed_type: multi-GPU
52
+ - gradient_accumulation_steps: 4
53
+ - total_train_batch_size: 16
54
+ - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
55
+ - lr_scheduler_type: cosine
56
+ - lr_scheduler_warmup_ratio: 0.1
57
+ - num_epochs: 1
58
+
59
+ ### Training results
60
+
61
+ | Training Loss | Epoch | Step | Validation Loss | Rewards/chosen | Rewards/rejected | Rewards/accuracies | Rewards/margins | Logps/rejected | Logps/chosen | Logits/rejected | Logits/chosen |
62
+ |:-------------:|:------:|:----:|:---------------:|:--------------:|:----------------:|:------------------:|:---------------:|:--------------:|:------------:|:---------------:|:-------------:|
63
+ | 0.693 | 0.0262 | 100 | 0.6929 | -0.0014 | -0.0019 | 0.5320 | 0.0006 | -297.2994 | -336.3557 | -3.1228 | -3.1361 |
64
+ | 0.6887 | 0.0523 | 200 | 0.6892 | -0.0302 | -0.0383 | 0.6160 | 0.0081 | -300.9348 | -339.2341 | -3.1215 | -3.1346 |
65
+ | 0.6789 | 0.0785 | 300 | 0.6794 | -0.0789 | -0.1087 | 0.6360 | 0.0299 | -307.9798 | -344.1051 | -3.1094 | -3.1216 |
66
+ | 0.6624 | 0.1047 | 400 | 0.6635 | -0.1807 | -0.2518 | 0.6390 | 0.0711 | -322.2854 | -354.2890 | -3.0664 | -3.0771 |
67
+ | 0.6373 | 0.1309 | 500 | 0.6503 | -0.2988 | -0.4120 | 0.6425 | 0.1133 | -338.3080 | -366.0959 | -2.9693 | -2.9839 |
68
+ | 0.6423 | 0.1570 | 600 | 0.6457 | -0.3891 | -0.5345 | 0.6375 | 0.1454 | -350.5518 | -375.1291 | -2.9372 | -2.9538 |
69
+ | 0.6266 | 0.1832 | 700 | 0.6420 | -0.7030 | -0.9081 | 0.6365 | 0.2051 | -387.9123 | -406.5211 | -2.9095 | -2.9229 |
70
+ | 0.5942 | 0.2094 | 800 | 0.6367 | -0.4969 | -0.6764 | 0.6475 | 0.1795 | -364.7484 | -385.9118 | -2.9255 | -2.9397 |
71
+ | 0.6171 | 0.2355 | 900 | 0.6330 | -0.5389 | -0.7443 | 0.6545 | 0.2054 | -371.5351 | -390.1065 | -2.8815 | -2.8992 |
72
+ | 0.6156 | 0.2617 | 1000 | 0.6271 | -0.9278 | -1.1788 | 0.6460 | 0.2510 | -414.9855 | -428.9975 | -2.8469 | -2.8665 |
73
+ | 0.6636 | 0.2879 | 1100 | 0.6234 | -0.7984 | -1.0304 | 0.6515 | 0.2320 | -400.1489 | -416.0618 | -2.8144 | -2.8347 |
74
+ | 0.6832 | 0.3141 | 1200 | 0.6152 | -1.0303 | -1.3170 | 0.6570 | 0.2866 | -428.8004 | -439.2536 | -2.7994 | -2.8212 |
75
+ | 0.5967 | 0.3402 | 1300 | 0.6131 | -1.2342 | -1.5321 | 0.6655 | 0.2979 | -450.3198 | -459.6400 | -2.7494 | -2.7756 |
76
+ | 0.596 | 0.3664 | 1400 | 0.6064 | -0.8587 | -1.1697 | 0.6820 | 0.3110 | -414.0766 | -422.0903 | -2.8084 | -2.8289 |
77
+ | 0.592 | 0.3926 | 1500 | 0.6027 | -0.9689 | -1.3189 | 0.6715 | 0.3499 | -428.9929 | -433.1132 | -2.7455 | -2.7703 |
78
+ | 0.6353 | 0.4187 | 1600 | 0.6051 | -0.9640 | -1.3223 | 0.6745 | 0.3582 | -429.3314 | -432.6226 | -2.6972 | -2.7245 |
79
+ | 0.6603 | 0.4449 | 1700 | 0.6016 | -0.9893 | -1.3221 | 0.6765 | 0.3328 | -429.3145 | -435.1521 | -2.7021 | -2.7305 |
80
+ | 0.5551 | 0.4711 | 1800 | 0.6023 | -1.0035 | -1.3765 | 0.6790 | 0.3731 | -434.7590 | -436.5641 | -2.6159 | -2.6492 |
81
+ | 0.5877 | 0.4973 | 1900 | 0.5975 | -0.8137 | -1.1853 | 0.6835 | 0.3716 | -415.6308 | -417.5872 | -2.6621 | -2.6941 |
82
+ | 0.5827 | 0.5234 | 2000 | 0.5935 | -0.8724 | -1.2562 | 0.6810 | 0.3838 | -422.7221 | -423.4575 | -2.6043 | -2.6396 |
83
+ | 0.6017 | 0.5496 | 2100 | 0.5911 | -1.0065 | -1.3971 | 0.6905 | 0.3907 | -436.8172 | -436.8658 | -2.6105 | -2.6436 |
84
+ | 0.5539 | 0.5758 | 2200 | 0.5920 | -0.9060 | -1.2945 | 0.6885 | 0.3884 | -426.5499 | -426.8195 | -2.5724 | -2.6076 |
85
+ | 0.5795 | 0.6019 | 2300 | 0.5914 | -1.1164 | -1.5398 | 0.6865 | 0.4234 | -451.0841 | -447.8605 | -2.5399 | -2.5757 |
86
+ | 0.5657 | 0.6281 | 2400 | 0.5904 | -1.0347 | -1.4494 | 0.6860 | 0.4147 | -442.0414 | -439.6861 | -2.5121 | -2.5487 |
87
+ | 0.5306 | 0.6543 | 2500 | 0.5918 | -1.0464 | -1.4840 | 0.6825 | 0.4376 | -445.5005 | -440.8591 | -2.4692 | -2.5102 |
88
+ | 0.5762 | 0.6805 | 2600 | 0.5927 | -1.0687 | -1.5141 | 0.6780 | 0.4455 | -448.5193 | -443.0862 | -2.4291 | -2.4735 |
89
+ | 0.6016 | 0.7066 | 2700 | 0.5936 | -1.0767 | -1.5080 | 0.6800 | 0.4313 | -447.9063 | -443.8889 | -2.4329 | -2.4747 |
90
+ | 0.6068 | 0.7328 | 2800 | 0.5897 | -1.1905 | -1.6433 | 0.6820 | 0.4527 | -461.4312 | -455.2722 | -2.4294 | -2.4708 |
91
+ | 0.5821 | 0.7590 | 2900 | 0.5870 | -1.1245 | -1.5598 | 0.6845 | 0.4353 | -453.0833 | -448.6697 | -2.4470 | -2.4862 |
92
+ | 0.5393 | 0.7851 | 3000 | 0.5873 | -1.2223 | -1.6710 | 0.6870 | 0.4486 | -464.2020 | -458.4521 | -2.4161 | -2.4565 |
93
+ | 0.577 | 0.8113 | 3100 | 0.5886 | -1.1359 | -1.5757 | 0.6845 | 0.4399 | -454.6796 | -449.8056 | -2.4137 | -2.4538 |
94
+ | 0.5731 | 0.8375 | 3200 | 0.5864 | -1.1928 | -1.6493 | 0.6900 | 0.4564 | -462.0313 | -455.5009 | -2.3988 | -2.4401 |
95
+ | 0.586 | 0.8636 | 3300 | 0.5865 | -1.1740 | -1.6231 | 0.6895 | 0.4492 | -459.4178 | -453.6159 | -2.3969 | -2.4384 |
96
+ | 0.5629 | 0.8898 | 3400 | 0.5860 | -1.1573 | -1.6086 | 0.6890 | 0.4513 | -457.9694 | -451.9486 | -2.3882 | -2.4306 |
97
+ | 0.6059 | 0.9160 | 3500 | 0.5858 | -1.1672 | -1.6213 | 0.6890 | 0.4541 | -459.2307 | -452.9388 | -2.3897 | -2.4320 |
98
+ | 0.5703 | 0.9422 | 3600 | 0.5860 | -1.1607 | -1.6138 | 0.6870 | 0.4532 | -458.4890 | -452.2865 | -2.3897 | -2.4320 |
99
+ | 0.5533 | 0.9683 | 3700 | 0.5858 | -1.1623 | -1.6161 | 0.6880 | 0.4538 | -458.7165 | -452.4510 | -2.3882 | -2.4304 |
100
+ | 0.5988 | 0.9945 | 3800 | 0.5862 | -1.1608 | -1.6138 | 0.6885 | 0.4530 | -458.4823 | -452.2973 | -2.3882 | -2.4306 |
101
+
102
+
103
+ ### Framework versions
104
+
105
+ - Transformers 4.41.1
106
+ - Pytorch 2.1.2
107
+ - Datasets 2.19.1
108
+ - Tokenizers 0.19.1
all_results.json ADDED
@@ -0,0 +1,9 @@
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "epoch": 1.0,
3
+ "total_flos": 0.0,
4
+ "train_loss": 0.6060711596530634,
5
+ "train_runtime": 35916.4658,
6
+ "train_samples": 61134,
7
+ "train_samples_per_second": 1.702,
8
+ "train_steps_per_second": 0.106
9
+ }
generation_config.json ADDED
@@ -0,0 +1,7 @@
 
 
 
 
 
 
 
 
1
+ {
2
+ "bos_token_id": 1,
3
+ "eos_token_id": 2,
4
+ "max_length": 2048,
5
+ "pad_token_id": 0,
6
+ "transformers_version": "4.41.1"
7
+ }
model.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:f81273e2e4858ec7753e12a1a02a023440ab4fb209b3625f8c365c66e91c1204
3
  size 4400216536
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:3792403f623109d4a40c553f4f959aba1949f3ba6457dda6779abaf241cf3fbf
3
  size 4400216536
runs/Jun02_10-42-23_poseidon/events.out.tfevents.1717325399.poseidon.2371184.0 CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:b72d835da0b4124b404817ee218908fbba50a3bc59d6eb7f9a999e95ee7a64bf
3
- size 295164
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:bdf5f8fc6ec2c96bebe381c84e67d771608094dfd93217d7ad0563800185383a
3
+ size 296894
train_results.json ADDED
@@ -0,0 +1,9 @@
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "epoch": 1.0,
3
+ "total_flos": 0.0,
4
+ "train_loss": 0.6060711596530634,
5
+ "train_runtime": 35916.4658,
6
+ "train_samples": 61134,
7
+ "train_samples_per_second": 1.702,
8
+ "train_steps_per_second": 0.106
9
+ }
trainer_state.json ADDED
The diff for this file is too large to render. See raw diff