Update README.md
Browse files
README.md
CHANGED
|
@@ -9,9 +9,22 @@ tags:
|
|
| 9 |
- llama
|
| 10 |
- trl
|
| 11 |
- sft
|
|
|
|
| 12 |
base_model: unsloth/tinyllama-chat-bnb-4bit
|
| 13 |
---
|
| 14 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 15 |
# Uploaded model
|
| 16 |
|
| 17 |
- **Developed by:** Ramikan-BR
|
|
|
|
| 9 |
- llama
|
| 10 |
- trl
|
| 11 |
- sft
|
| 12 |
+
- datasets [Ramikan-BR/code.evol.instruct.wiz.oss_python.json](https://huggingface.co/datasets/Ramikan-BR/code.evol.instruct.wiz.oss_python.json)
|
| 13 |
base_model: unsloth/tinyllama-chat-bnb-4bit
|
| 14 |
---
|
| 15 |
|
| 16 |
+
datasets: [Ramikan-BR/code.evol.instruct.wiz.oss_python.json](https://huggingface.co/datasets/Ramikan-BR/code.evol.instruct.wiz.oss_python.json)
|
| 17 |
+
```python
|
| 18 |
+
==((====))== Unsloth - 2x faster free finetuning | Num GPUs = 1
|
| 19 |
+
\\ /| Num examples = 937 | Num Epochs = 2
|
| 20 |
+
O^O/ \_/ \ Batch size per device = 2 | Gradient Accumulation steps = 256
|
| 21 |
+
\ / Total batch size = 512 | Total steps = 2
|
| 22 |
+
"-____-" Number of trainable parameters = 201,850,880
|
| 23 |
+
[2/2 22:36, Epoch 1/2]
|
| 24 |
+
Step Training Loss
|
| 25 |
+
1 0.707400
|
| 26 |
+
2 0.717800
|
| 27 |
+
|
| 28 |
# Uploaded model
|
| 29 |
|
| 30 |
- **Developed by:** Ramikan-BR
|