Updating evaluation details for RULER (reasoning off)
#6
by
ameyasunilm
- opened
README.md
CHANGED
|
@@ -63,7 +63,7 @@ GOVERNING TERMS: This trial service is governed by the [NVIDIA API Trial Terms o
|
|
| 63 |
|
| 64 |
### Benchmark Results (Reasoning On)
|
| 65 |
|
| 66 |
-
We evaluated our model in
|
| 67 |
|
| 68 |
|
| 69 |
| Benchmark | Qwen3-8B | NVIDIA-Nemotron-Nano-9B-v2 |
|
|
|
|
| 63 |
|
| 64 |
### Benchmark Results (Reasoning On)
|
| 65 |
|
| 66 |
+
We evaluated our model in **Reasoning-On** mode across all benchmarks, except RULER, which is evaluated in **Reasoning-Off** mode.
|
| 67 |
|
| 68 |
|
| 69 |
| Benchmark | Qwen3-8B | NVIDIA-Nemotron-Nano-9B-v2 |
|