Sponsored by
Vmake Labs Video Enhancer
-Vmake Labs Video Enhancer uses AI to improve video quality, restore details, reduce blur, and upscale videos to 8K.
📚
Part of the
Aether Foundation Model collection
— base, instruction-tuned, and checkpoints in one place.
Instruction-tuned (SFT) version of the fully-open
Aether-7B-5Attn
base model.
Post-trained for multiple-choice / benchmark-style answering. 6.59B MoE (~2.98B active), 49 layers on a 7×7 Latin square.
Apache-2.0.
Learning rate chosen by held-out accuracy, not loss
Full-parameter SFT was run at three learning rates (2e-6 / 6e-6 / 2e-5). The final checkpoint was selected by
held-out benchmark accuracy, not training loss
— for small models a high LR lowers training loss while degrading real capability, so loss is the wrong selector.
vs base (before SFT):
base 26.7% →
34.9% (+8.2pp)
— same held-out set, same harness.
All evaluations use held-out sets not seen in training. SFT was performed on MMLU-auxiliary (multiple-choice format), which does not overlap with the evaluation subjects (GPQA / K-AI).
Pretraining data mix (inherited from base)
Domain
Share
Math (finemath + open-web-math)
37.8%
Korean (webtext + synth)
21.6%
English web & synthetic (fineweb-edu + cosmopedia)
21.6%
Code (opc)
13.5%
phase15 pre-blend
5.4%
This is the
base pretraining
mix. This model's post-training (SFT) data is MMLU-auxiliary.
Architecture (inherited from base)
Identical to Aether-7B-5Attn base: 49 layers placed on a
7×7 Latin square
with heterogeneous attention (7 labels / 5 distinct mechanisms) and a 25-expert MoE (top-7 + 1 shared). Full structural detail and the diagram are in the base card, §3.2:
FINAL-Bench/Aether-7B-5Attn
.
Relative to the base Aether. Among six sovereign fully-open models, VIDRAFT is the only single AI startup, and Aether has the most attention types (5) in a Latin-square layout.
Usage
Uses the custom
aether_pkg/
architecture (
trust_remote_code=True
). Follow the loading instructions in the
base model card
.
⚠️
Use batch_size=1.
Some attention branches (NSA family) do not consume a padding mask, so batching padded sequences can silently corrupt results. Run inference one sequence at a time.
Aether-7B-5Attn-it huggingface.co is an AI model on huggingface.co that provides Aether-7B-5Attn-it's model effect (), which can be used instantly with this FINAL-Bench Aether-7B-5Attn-it model. huggingface.co supports a free trial of the Aether-7B-5Attn-it model, and also provides paid use of the Aether-7B-5Attn-it. Support call Aether-7B-5Attn-it model through api, including Node.js, Python, http.
Aether-7B-5Attn-it huggingface.co is an online trial and call api platform, which integrates Aether-7B-5Attn-it's modeling effects, including api services, and provides a free online trial of Aether-7B-5Attn-it, you can try Aether-7B-5Attn-it online for free by clicking the link below.
FINAL-Bench Aether-7B-5Attn-it online free url in huggingface.co:
Aether-7B-5Attn-it is an open source model from GitHub that offers a free installation service, and any user can find Aether-7B-5Attn-it on GitHub to install. At the same time, huggingface.co provides the effect of Aether-7B-5Attn-it install, users can directly use Aether-7B-5Attn-it installed effect in huggingface.co for debugging and trial. It also supports api for free installation.