Update README.md
Browse files
README.md
CHANGED
|
@@ -27,59 +27,4 @@ model-index:
|
|
| 27 |
- name: smollm2-1.7B-8k-mix7-ep2-v2
|
| 28 |
results: []
|
| 29 |
---
|
| 30 |
-
|
| 31 |
-
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
|
| 32 |
-
should probably proofread and complete it, then remove this comment. -->
|
| 33 |
-
|
| 34 |
-
[<img src="https://raw.githubusercontent.com/wandb/assets/main/wandb-github-badge-28.svg" alt="Visualize in Weights & Biases" width="200" height="32"/>](https://wandb.ai/loubnabnl/huggingface/runs/6rp7tpcv)
|
| 35 |
-
# smollm2-1.7B-8k-mix7-ep2-v2
|
| 36 |
-
|
| 37 |
-
This model is a fine-tuned version of [HuggingFaceTB/SmolLM2-1.7B-8k](https://huggingface.co/HuggingFaceTB/SmolLM2-1.7B-8k) on the HuggingFaceTB/magpie-ultra-v1.0-filtered-400K-H4, the HuggingFaceTB/OpenHermes-2.5-H4-200k, the HuggingFaceTB/ifeval-like-data-36k-H4, the HuggingFaceTB/Numina-CoT-H4, the HuggingFaceTB/MetaMathQA-H4-200k, the HuggingFaceTB/self-oss-instruct-sc2-H4, the HuggingFaceTB/systemchats2.0-H4-short, the HuggingFaceTB/summarization-data-10k-H4, the HuggingFaceTB/everyday-conversations-llama3.1-2k, the HuggingFaceTB/apigen-smollm-trl-FC-H4, the HuggingFaceTB/smollm-v2-summarization, the HuggingFaceTB/smollm-v2-rewriting-50k-H4, the HuggingFaceTB/explore-instruct-rewrite-H4 and the HuggingFaceTB/LongAlign-16k-ctx-english-H4 datasets.
|
| 38 |
-
It achieves the following results on the evaluation set:
|
| 39 |
-
- Loss: 1.0630
|
| 40 |
-
|
| 41 |
-
## Model description
|
| 42 |
-
|
| 43 |
-
More information needed
|
| 44 |
-
|
| 45 |
-
## Intended uses & limitations
|
| 46 |
-
|
| 47 |
-
More information needed
|
| 48 |
-
|
| 49 |
-
## Training and evaluation data
|
| 50 |
-
|
| 51 |
-
More information needed
|
| 52 |
-
|
| 53 |
-
## Training procedure
|
| 54 |
-
|
| 55 |
-
### Training hyperparameters
|
| 56 |
-
|
| 57 |
-
The following hyperparameters were used during training:
|
| 58 |
-
- learning_rate: 0.0003
|
| 59 |
-
- train_batch_size: 4
|
| 60 |
-
- eval_batch_size: 4
|
| 61 |
-
- seed: 42
|
| 62 |
-
- distributed_type: multi-GPU
|
| 63 |
-
- num_devices: 8
|
| 64 |
-
- gradient_accumulation_steps: 4
|
| 65 |
-
- total_train_batch_size: 128
|
| 66 |
-
- total_eval_batch_size: 32
|
| 67 |
-
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
|
| 68 |
-
- lr_scheduler_type: cosine
|
| 69 |
-
- lr_scheduler_warmup_ratio: 0.1
|
| 70 |
-
- num_epochs: 2
|
| 71 |
-
|
| 72 |
-
### Training results
|
| 73 |
-
|
| 74 |
-
| Training Loss | Epoch | Step | Validation Loss |
|
| 75 |
-
|:-------------:|:------:|:----:|:---------------:|
|
| 76 |
-
| 0.7442 | 0.9992 | 893 | 1.0789 |
|
| 77 |
-
| 0.6705 | 1.9983 | 1786 | 1.0630 |
|
| 78 |
-
|
| 79 |
-
|
| 80 |
-
### Framework versions
|
| 81 |
-
|
| 82 |
-
- Transformers 4.42.3
|
| 83 |
-
- Pytorch 2.1.2
|
| 84 |
-
- Datasets 2.20.0
|
| 85 |
-
- Tokenizers 0.19.1
|
|
|
|
| 27 |
- name: smollm2-1.7B-8k-mix7-ep2-v2
|
| 28 |
results: []
|
| 29 |
---
|
| 30 |
+
SFT only version of https://huggingface.co/HuggingFaceTB/SmolLM2-1.7B-Instruct
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|