uploaded readme
Browse files
README.md
ADDED
|
@@ -0,0 +1,100 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
Quantization made by Richard Erkhov.
|
| 2 |
+
|
| 3 |
+
[Github](https://github.com/RichardErkhov)
|
| 4 |
+
|
| 5 |
+
[Discord](https://discord.gg/pvy7H8DZMG)
|
| 6 |
+
|
| 7 |
+
[Request more models](https://github.com/RichardErkhov/quant_request)
|
| 8 |
+
|
| 9 |
+
|
| 10 |
+
mpt_1000_STEPS_1e6_SFT_SFT - bnb 8bits
|
| 11 |
+
- Model creator: https://huggingface.co/tsavage68/
|
| 12 |
+
- Original model: https://huggingface.co/tsavage68/mpt_1000_STEPS_1e6_SFT_SFT/
|
| 13 |
+
|
| 14 |
+
|
| 15 |
+
|
| 16 |
+
|
| 17 |
+
Original model description:
|
| 18 |
+
---
|
| 19 |
+
license: apache-2.0
|
| 20 |
+
base_model: mosaicml/mpt-7b-instruct
|
| 21 |
+
tags:
|
| 22 |
+
- trl
|
| 23 |
+
- sft
|
| 24 |
+
- generated_from_trainer
|
| 25 |
+
model-index:
|
| 26 |
+
- name: mpt_1000_STEPS_1e6_SFT_SFT
|
| 27 |
+
results: []
|
| 28 |
+
---
|
| 29 |
+
|
| 30 |
+
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
|
| 31 |
+
should probably proofread and complete it, then remove this comment. -->
|
| 32 |
+
|
| 33 |
+
# mpt_1000_STEPS_1e6_SFT_SFT
|
| 34 |
+
|
| 35 |
+
This model is a fine-tuned version of [mosaicml/mpt-7b-instruct](https://huggingface.co/mosaicml/mpt-7b-instruct) on an unknown dataset.
|
| 36 |
+
It achieves the following results on the evaluation set:
|
| 37 |
+
- Loss: 0.4128
|
| 38 |
+
|
| 39 |
+
## Model description
|
| 40 |
+
|
| 41 |
+
More information needed
|
| 42 |
+
|
| 43 |
+
## Intended uses & limitations
|
| 44 |
+
|
| 45 |
+
More information needed
|
| 46 |
+
|
| 47 |
+
## Training and evaluation data
|
| 48 |
+
|
| 49 |
+
More information needed
|
| 50 |
+
|
| 51 |
+
## Training procedure
|
| 52 |
+
|
| 53 |
+
### Training hyperparameters
|
| 54 |
+
|
| 55 |
+
The following hyperparameters were used during training:
|
| 56 |
+
- learning_rate: 1e-06
|
| 57 |
+
- train_batch_size: 2
|
| 58 |
+
- eval_batch_size: 1
|
| 59 |
+
- seed: 42
|
| 60 |
+
- gradient_accumulation_steps: 2
|
| 61 |
+
- total_train_batch_size: 4
|
| 62 |
+
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
|
| 63 |
+
- lr_scheduler_type: cosine
|
| 64 |
+
- lr_scheduler_warmup_steps: 100
|
| 65 |
+
- training_steps: 1000
|
| 66 |
+
|
| 67 |
+
### Training results
|
| 68 |
+
|
| 69 |
+
| Training Loss | Epoch | Step | Validation Loss |
|
| 70 |
+
|:-------------:|:-----:|:----:|:---------------:|
|
| 71 |
+
| 1.568 | 0.05 | 50 | 1.4778 |
|
| 72 |
+
| 0.4601 | 0.1 | 100 | 0.4773 |
|
| 73 |
+
| 0.4655 | 0.15 | 150 | 0.4432 |
|
| 74 |
+
| 0.4866 | 0.2 | 200 | 0.4338 |
|
| 75 |
+
| 0.4309 | 0.24 | 250 | 0.4279 |
|
| 76 |
+
| 0.4481 | 0.29 | 300 | 0.4238 |
|
| 77 |
+
| 0.4239 | 0.34 | 350 | 0.4206 |
|
| 78 |
+
| 0.4025 | 0.39 | 400 | 0.4184 |
|
| 79 |
+
| 0.4377 | 0.44 | 450 | 0.4169 |
|
| 80 |
+
| 0.4192 | 0.49 | 500 | 0.4154 |
|
| 81 |
+
| 0.407 | 0.54 | 550 | 0.4145 |
|
| 82 |
+
| 0.4291 | 0.59 | 600 | 0.4136 |
|
| 83 |
+
| 0.4048 | 0.64 | 650 | 0.4133 |
|
| 84 |
+
| 0.397 | 0.68 | 700 | 0.4131 |
|
| 85 |
+
| 0.4016 | 0.73 | 750 | 0.4128 |
|
| 86 |
+
| 0.4108 | 0.78 | 800 | 0.4128 |
|
| 87 |
+
| 0.4427 | 0.83 | 850 | 0.4128 |
|
| 88 |
+
| 0.3882 | 0.88 | 900 | 0.4127 |
|
| 89 |
+
| 0.3929 | 0.93 | 950 | 0.4127 |
|
| 90 |
+
| 0.4125 | 0.98 | 1000 | 0.4128 |
|
| 91 |
+
|
| 92 |
+
|
| 93 |
+
### Framework versions
|
| 94 |
+
|
| 95 |
+
- Transformers 4.39.3
|
| 96 |
+
- Pytorch 2.0.0+cu117
|
| 97 |
+
- Datasets 2.18.0
|
| 98 |
+
- Tokenizers 0.15.2
|
| 99 |
+
|
| 100 |
+
|