Update README.md
Browse files
README.md
CHANGED
|
@@ -46,6 +46,14 @@ tags:
|
|
| 46 |
|
| 47 |
Introducing **void**: our first ever language model, trained from scratch with a novel hybrid tokenizer on 300M high-quality tokens (total of 2 epochs on a B300) using 4096 as context window. Total cost was $23 dollars. It has around 140m parameters. Future releases are expected to be published in the following weeks.
|
| 48 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 49 |
**Disclaimer:** Even though the model is based on gemma 3 architecture, the tokenizer is different so you might need to wait until this model can be added to llama.cpp
|
| 50 |
|
| 51 |
If you want to sponsor future model releases, you can get information on how to make contributions here: [CEAMFA](https://huggingface.co/CEAMFA)
|
|
|
|
| 46 |
|
| 47 |
Introducing **void**: our first ever language model, trained from scratch with a novel hybrid tokenizer on 300M high-quality tokens (total of 2 epochs on a B300) using 4096 as context window. Total cost was $23 dollars. It has around 140m parameters. Future releases are expected to be published in the following weeks.
|
| 48 |
|
| 49 |
+
| Benchmark | Accuracy | Normalized |
|
| 50 |
+
| ------------- | -------: | ---------: |
|
| 51 |
+
| ARC Challenge | 25.17% | 27.22% |
|
| 52 |
+
| ARC Easy | 45.03% | 43.01% |
|
| 53 |
+
| HellaSwag | 31.57% | 35.77% |
|
| 54 |
+
| PIQA | 61.15% | 60.12% |
|
| 55 |
+
| WinoGrande | 53.12% | — |
|
| 56 |
+
|
| 57 |
**Disclaimer:** Even though the model is based on gemma 3 architecture, the tokenizer is different so you might need to wait until this model can be added to llama.cpp
|
| 58 |
|
| 59 |
If you want to sponsor future model releases, you can get information on how to make contributions here: [CEAMFA](https://huggingface.co/CEAMFA)
|