Model comparison
ERNIE 5.0 0110 vs Phi-2
ERNIE 5.0 0110 has enough public results to be ranked (#129); Phi-2 does not yet, so treat this comparison as directional.
Last verified . 0 shared benchmarks.
Baidu
MicrosoftSummary
- Phi-2 has downloadable open weights; the other is API-only.
Side by side
| ERNIE 5.0 0110 | Phi-2 | |
|---|---|---|
| Provider | Baidu | Microsoft |
| Noometry Index | 41.8 | — |
| Released | — | 2023-12-12 |
| Weights | Proprietary | Open |
| Context window | — | — |
| Max output | — | — |
| Input $ / M tokens | — | — |
| Output $ / M tokens | — | — |
| Results tracked | 20 | 11 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Not comparable
ERNIE 5.0 0110: 43.0 (#94), Phi-2: —
| Benchmark | ERNIE 5.0 0110 | Phi-2 |
|---|---|---|
| LMArena Coding | 1455 | — |
| HumanEval+ | — | 45.1% |
| MBPP+ | — | 54.2% |
Reasoning Not comparable
ERNIE 5.0 0110: 17.0 (#297), Phi-2: —
| Benchmark | ERNIE 5.0 0110 | Phi-2 |
|---|---|---|
| NYT Connections (extended) | 10.3% | — |
| Thematic Generalization | 41.7% | — |
| LMArena Hard Prompts | 1445 | — |
| Adversarial NLI | — | 42.5% |
| BIG-Bench Hard | — | 59.4% |
| Epoch Capabilities Index | — | 107.94 |
| HellaSwag | — | 53.6% |
| WinoGrande | — | 54.7% |
Math Not comparable
ERNIE 5.0 0110: 39.3 (#110), Phi-2: —
| Benchmark | ERNIE 5.0 0110 | Phi-2 |
|---|---|---|
| LMArena Math | 1437 | — |
Knowledge Not comparable
ERNIE 5.0 0110: 39.8 (#128), Phi-2: —
| Benchmark | ERNIE 5.0 0110 | Phi-2 |
|---|---|---|
| LMArena Expert | 1428 | — |
| ARC (AI2) Challenge | — | 75.9% |
| MMLU | — | 58.4% |
| OpenBookQA | — | 73.6% |
| TriviaQA | — | 45.2% |
Multimodal Not comparable
ERNIE 5.0 0110: 39.9 (#53), Phi-2: —
| Benchmark | ERNIE 5.0 0110 | Phi-2 |
|---|---|---|
| LMArena Vision | 1249 | — |
Multilingual Not comparable
ERNIE 5.0 0110: 54.1 (#49), Phi-2: —
| Benchmark | ERNIE 5.0 0110 | Phi-2 |
|---|---|---|
| LMArena Non-English | 1436 | — |
| LMArena Chinese | 1512 | — |
| LMArena French | 1467 | — |
| LMArena German | 1460 | — |
| LMArena Japanese | 1382 | — |
| LMArena Korean | 1406 | — |
| LMArena Russian | 1446 | — |
| LMArena Spanish | 1473 | — |
Instruction Following Not comparable
ERNIE 5.0 0110: 74.5 (#92), Phi-2: —
| Benchmark | ERNIE 5.0 0110 | Phi-2 |
|---|---|---|
| LMArena Instruction Following | 1413 | — |
Long Context Not comparable
ERNIE 5.0 0110: 43.4 (#95), Phi-2: —
| Benchmark | ERNIE 5.0 0110 | Phi-2 |
|---|---|---|
| LMArena Longer Query | 1422 | — |
Writing & Preference Not comparable
ERNIE 5.0 0110: 63.1 (#66), Phi-2: —
| Benchmark | ERNIE 5.0 0110 | Phi-2 |
|---|---|---|
| LMArena Text | 1445 | — |
| LMArena Creative Writing | 1426 | — |
| LMArena Multi-Turn | 1434 | — |
Frequently asked questions
Is ERNIE 5.0 0110 better than Phi-2?
ERNIE 5.0 0110 has enough public results to be ranked (#129); Phi-2 does not yet, so treat this comparison as directional.
How many benchmarks do ERNIE 5.0 0110 and Phi-2 share?
0 benchmarks have published results for both models. ERNIE 5.0 0110 has 20 scored results on Noometry and Phi-2 has 11.