Model comparison
Nemotron-4 15B vs Phi-2
Neither Nemotron-4 15B nor Phi-2 has enough public benchmark results to be ranked yet; the rows below show what has been published.
Last verified . 6 shared benchmarks.
NVIDIA
MicrosoftSummary
- They share 6 benchmarks with published results for both.
- Phi-2 has downloadable open weights; the other is API-only.
Side by side
| Nemotron-4 15B | Phi-2 | |
|---|---|---|
| Provider | NVIDIA | Microsoft |
| Noometry Index | — | — |
| Released | 2024-02-26 | 2023-12-12 |
| Weights | Proprietary | Open |
| Context window | — | — |
| Max output | — | — |
| Input $ / M tokens | — | — |
| Output $ / M tokens | — | — |
| Results tracked | 8 | 11 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Not comparable
Nemotron-4 15B: —, Phi-2: —
| Benchmark | Nemotron-4 15B | Phi-2 |
|---|---|---|
| HumanEval+ | — | 45.1% |
| MBPP+ | — | 54.2% |
Reasoning Not comparable
Nemotron-4 15B: —, Phi-2: —
| Benchmark | Nemotron-4 15B | Phi-2 |
|---|---|---|
| BIG-Bench Hard | 58.7% | 59.4% |
| Epoch Capabilities Index | 107.72 | 107.94 |
| HellaSwag | 82.4% | 53.6% |
| WinoGrande | 78% | 54.7% |
| Adversarial NLI | — | 42.5% |
| PIQA | 82.4% | — |
Math Not comparable
Nemotron-4 15B: —, Phi-2: —
| Benchmark | Nemotron-4 15B | Phi-2 |
|---|---|---|
| GSM8K | 46% | — |
Knowledge Not comparable
Nemotron-4 15B: —, Phi-2: —
| Benchmark | Nemotron-4 15B | Phi-2 |
|---|---|---|
| ARC (AI2) Challenge | 55.5% | 75.9% |
| MMLU | 58.7% | 58.4% |
| OpenBookQA | — | 73.6% |
| TriviaQA | — | 45.2% |
Frequently asked questions
Is Nemotron-4 15B better than Phi-2?
Neither Nemotron-4 15B nor Phi-2 has enough public benchmark results to be ranked yet; the rows below show what has been published.
How many benchmarks do Nemotron-4 15B and Phi-2 share?
6 benchmarks have published results for both models. Nemotron-4 15B has 8 scored results on Noometry and Phi-2 has 11.