Update README.md
Browse files
README.md
CHANGED
|
@@ -9,6 +9,14 @@ base_model:
|
|
| 9 |
- nvidia/Llama-3.1-Nemotron-Nano-8B-v1
|
| 10 |
---
|
| 11 |
# Model Overview
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 12 |
|
| 13 |
### Description:
|
| 14 |
DLER-Llama-3.1-Nemotron-8B is an ultra-efficient 8B open-weight reasoning model designed for challenging tasks such as mathematics, programming, and scientific problem-solving. It is first trained with the DLER algorithm on agentica-org/DeepScaleR-Preview-Dataset and then enhanced using a weight-merging technique to merge with the base model to mitigate accuracy degradation. Compared to the Llama-3.1-Nemotron-8B model, DLER-Llama-Nemotron-8B-Merge achieves substantial efficiency gains, reducing the average response length by nearly 50% across diverse mathematical benchmarks without sacrificing accuracy.
|
|
|
|
| 9 |
- nvidia/Llama-3.1-Nemotron-Nano-8B-v1
|
| 10 |
---
|
| 11 |
# Model Overview
|
| 12 |
+
<div align="center">
|
| 13 |
+
<span style="font-family: default; font-size: 1.5em;">DLER-Llama-Nemotron-8B-Merge</span>
|
| 14 |
+
<div>
|
| 15 |
+
🚀 The leading efficient reasoning model for cutting-edge research and development 🌟
|
| 16 |
+
</div>
|
| 17 |
+
</div>
|
| 18 |
+
|
| 19 |
+
|
| 20 |
|
| 21 |
### Description:
|
| 22 |
DLER-Llama-3.1-Nemotron-8B is an ultra-efficient 8B open-weight reasoning model designed for challenging tasks such as mathematics, programming, and scientific problem-solving. It is first trained with the DLER algorithm on agentica-org/DeepScaleR-Preview-Dataset and then enhanced using a weight-merging technique to merge with the base model to mitigate accuracy degradation. Compared to the Llama-3.1-Nemotron-8B model, DLER-Llama-Nemotron-8B-Merge achieves substantial efficiency gains, reducing the average response length by nearly 50% across diverse mathematical benchmarks without sacrificing accuracy.
|