Instructions to use Boffl/BullingerLM-llama3.1-70B-pretrain-EA with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- PEFT
How to use Boffl/BullingerLM-llama3.1-70B-pretrain-EA with PEFT:
from peft import PeftModel from transformers import AutoModelForCausalLM base_model = AutoModelForCausalLM.from_pretrained("unsloth/Meta-Llama-3.1-70B-bnb-4bit") model = PeftModel.from_pretrained(base_model, "Boffl/BullingerLM-llama3.1-70B-pretrain-EA") - Notebooks
- Google Colab
- Kaggle
Upload all_results.json with huggingface_hub
Browse files- all_results.json +8 -0
all_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 0.9993906154783668,
|
| 3 |
+
"total_flos": 5.610936315118879e+18,
|
| 4 |
+
"train_loss": 2.5786649890062288,
|
| 5 |
+
"train_runtime": 34559.7112,
|
| 6 |
+
"train_samples_per_second": 0.38,
|
| 7 |
+
"train_steps_per_second": 0.012
|
| 8 |
+
}
|