arco lite is a passthrough arco model based on danube outputs to keep generality, even though its performance decreased, it's stil competitive to qwen2 at most benchmarks, being mmlu the only reason why is better on average. Note, arco-lite is still un-trained, i'm expecting it to be better after some iterations.
benchmarks
zero-shot evaluations, as you can see is smarter than qwen but without world knowledge, so don't use it for tasks that need factual output.
Parameters
Model
MMLU
ARC
HellaSwag
PIQA
Winogrande
Average
488m
arco-lite
23.22
33.45
56.55
69.70
59.19
48.46
494m
qwen2
44.13
28.92
49.05
69.31
56.99
49.68
Configuration
The following YAML configuration was used to produce this model:
More Information About arco-lite huggingface.co Model
arco-lite huggingface.co
arco-lite huggingface.co is an AI model on huggingface.co that provides arco-lite's model effect (), which can be used instantly with this appvoid arco-lite model. huggingface.co supports a free trial of the arco-lite model, and also provides paid use of the arco-lite. Support call arco-lite model through api, including Node.js, Python, http.
arco-lite huggingface.co is an online trial and call api platform, which integrates arco-lite's modeling effects, including api services, and provides a free online trial of arco-lite, you can try arco-lite online for free by clicking the link below.
appvoid arco-lite online free url in huggingface.co:
arco-lite is an open source model from GitHub that offers a free installation service, and any user can find arco-lite on GitHub to install. At the same time, huggingface.co provides the effect of arco-lite install, users can directly use arco-lite installed effect in huggingface.co for debugging and trial. It also supports api for free installation.