MetaNova-0.1-70M

WhirlwindAI/MetaNova-0.1-70M

MetaNova-0.1-70M is a text generation model from WhirlwindAI. Use it when you need the model to write or continue text. It is set up for transformers. The card lists the license as apache-2.0.

Downloads · 30 days

223

77% of all-time downloads

All-time downloads

290

Public

Parameters

70.5M

141 MB on disk

Likes

Public

Hugging Face

Repo makeup

Click a slice to open those files.

.safetensors141 MB · 98%

At a glance

Task: Text Generation
Library: transformers
License: apache-2.0
Model type: metanova_run2
Access: Public
Created: Jul 2, 2026
Updated: Sep 29, 2026
SHA: ce7d88a0

Try a prompt

Model card

From the Hugging Face model README

</div> <img src="https://user-images.githubusercontent.com/74038190/212284100-561aa473-3905-4a80-b561-0d28506553ee.gif" width="100%">

⚡ At a Glance

<table> <thead> <tr> <th align="center" style="background:rgba(139,92,246,0.18);color:#c4b5fd;border-bottom:2px solid #8b5cf6;">Model</th> <th align="center" style="background:rgba(139,92,246,0.18);color:#c4b5fd;border-bottom:2px solid #8b5cf6;">Params</th> <th align="center" style="background:rgba(139,92,246,0.18);color:#c4b5fd;border-bottom:2px solid #8b5cf6;">Final Avg</th> <th align="center" style="background:rgba(139,92,246,0.18);color:#c4b5fd;border-bottom:2px solid #8b5cf6;">Highlights</th> </tr> </thead> <tbody> <tr> <td align="center" style="background:rgba(167,139,250,0.08);"> <span style="color:#a78bfa;"><b>WhirlwindAI/MetaNova-Test</b></span> </td> <td align="center">70.55M</td> <td align="center"><span style="color:#c4b5fd;"><b>31.94%</b></span></td> <td align="center"><span style="color:#c4b5fd;">🏅 ARC-Challenge (25.09%)</span></td> </tr> <tr> <td align="center" style="background:rgba(96,165,250,0.08);"> <span style="color:#60a5fa;"><b>WhirlwindAI/MetaNova-0.1-70M</b></span> </td> <td align="center">70.55M</td> <td align="center"><span style="color:#93c5fd;"><b>35.36%</b></span></td> <td align="center"><span style="color:#93c5fd;">🏅 ArithMark-2 (27.12%)</span></td> </tr> <tr> <td align="center" style="background:rgba(52,211,153,0.08);"> <span style="color:#34d399;"><b>SupraLabs/Supra-50M-Base</b></span> </td> <td align="center"><span style="color:#fbbf24;">51.79M</span></td> <td align="center"><span style="color:#34d399;"><b>38.60%</b></span></td> <td align="center"><span style="color:#34d399;">🏅 HellaSwag, ARC-Easy, PIQA, ARC Avg, Final Avg</span></td> </tr> </tbody> </table> </div>

🟣 <span style="color:#c4b5fd;">Quick summary</span> — The WhirlwindAI MetaNova models each have their own strengths: MetaNova-0.1-70M leads on ArithMark-2, while MetaNova-Test takes ARC-Challenge. Supra-50M-Base (external baseline) posts the highest Final Avg at 38.60% with the smallest parameter count.

🏁 Full Leaderboard

<div align="center"> <table> <thead> <tr> <th align="center" style="background:rgba(139,92,246,0.18);color:#c4b5fd;border-bottom:2px solid #8b5cf6;">Benchmark</th> <th align="center" style="background:rgba(139,92,246,0.18);color:#a78bfa;border-bottom:2px solid #8b5cf6;">WhirlwindAI/<br/>MetaNova-Test</th> <th align="center" style="background:rgba(139,92,246,0.18);color:#93c5fd;border-bottom:2px solid #8b5cf6;">WhirlwindAI/<br/>MetaNova-0.1-BF16</th> <th align="center" style="background:rgba(139,92,246,0.18);color:#6ee7b7;border-bottom:2px solid #8b5cf6;">SupraLabs/<br/>Supra-50M-Base</th> </tr> </thead> <tbody> <tr> <td align="center" style="background:rgba(139,92,246,0.10);color:#e9d5ff;"><b>Params</b></td> <td align="center">70.55M</td> <td align="center">70.55M</td> <td align="center"><span style="color:#fbbf24;"><b>51.79M</b></span></td> </tr> <tr> <td align="center" style="background:rgba(139,92,246,0.10);color:#e9d5ff;">HellaSwag</td> <td align="center"><span style="color:#a1a1aa;">25.37%</span></td> <td align="center"><span style="color:#a1a1aa;">27.40%</span></td> <td align="center"><span style="color:#34d399;"><b>31.65%</b></span></td> </tr> <tr> <td align="center" style="background:rgba(139,92,246,0.10);color:#e9d5ff;">ARC-Easy</td> <td align="center"><span style="color:#a1a1aa;">26.89%</span></td> <td align="center"><span style="color:#a1a1aa;">31.73%</span></td> <td align="center"><span style="color:#34d399;"><b>45.58%</b></span></td> </tr> <tr> <td align="center" style="background:rgba(139,92,246,0.10);color:#e9d5ff;">ARC-Challenge</td> <td align="center"><span style="color:#34d399;"><b>25.09%</b></span></td> <td align="center"><span style="color:#a1a1aa;">25.00%</span></td> <td align="center"><span style="color:#a1a1aa;">24.66%</span></td> </tr> <tr> <td align="center" style="background:rgba(139,92,246,0.10);color:#e9d5ff;">PIQA</td> <td align="center"><span style="color:#a1a1aa;">52.29%</span></td> <td align="center"><span style="color:#a1a1aa;">58.54%</span></td> <td align="center"><span style="color:#34d399;"><b>61.53%</b></span></td> </tr> <tr> <td align="center" style="background:rgba(139,92,246,0.10);color:#e9d5ff;">ArithMark-2</td> <td align="center"><span style="color:#a1a1aa;">24.12%</span></td> <td align="center"><span style="color:#34d399;"><b>27.12%</b></span></td> <td align="center"><span style="color:#a1a1aa;">26.08%</span></td> </tr> <tr> <td align="center" style="background:rgba(139,92,246,0.10);color:#e9d5ff;">ARC Avg</td> <td align="center"><span style="color:#a1a1aa;">25.99%</span></td> <td align="center"><span style="color:#a1a1aa;">28.37%</span></td> <td align="center"><span style="color:#34d399;"><b>35.12%</b></span></td> </tr> <tr> <td align="center" style="background:rgba(139,92,246,0.22);color:#ddd6fe;border-top:2px solid #8b5cf6;"><b>Final Avg</b></td> <td align="center" style="border-top:2px solid #8b5cf6;"><span style="color:#c4b5fd;"><b>31.94%</b></span></td> <td align="center" style="border-top:2px solid #8b5cf6;"><span style="color:#93c5fd;"><b>35.36%</b></span></td> <td align="center" style="border-top:2px solid #8b5cf6;"><span style="color:#34d399;"><b>38.60%</b></span> 🏆</td> </tr> </tbody> </table>

<sub>🏆 = best score in the row. <b>Bold</b> = row winner.</sub>

</div>

📊 Metric-by-Metric View

<details open> <summary><b>🟣 HellaSwag</b></summary>

Model	Score
<span style="color:#a78bfa;">MetaNova-Test</span>	25.37%	<span style="color:#a78bfa;">█████</span><span style="color:#3f3f46;">░░░░░░░</span>
<span style="color:#60a5fa;">MetaNova-0.1-BF16</span>	27.40%	<span style="color:#60a5fa;">█████</span><span style="color:#3f3f46;">░░░░░░░</span>
<span style="color:#34d399;">Supra-50M-Base</span>	31.65%	<span style="color:#34d399;">██████</span><span style="color:#3f3f46;">░░░░░░</span>

</details> <details open> <summary><b>🟣 ARC-Easy</b></summary>

Model	Score
<span style="color:#a78bfa;">MetaNova-Test</span>	26.89%	<span style="color:#a78bfa;">█████</span><span style="color:#3f3f46;">░░░░░░░</span>
<span style="color:#60a5fa;">MetaNova-0.1-BF16</span>	31.73%	<span style="color:#60a5fa;">██████</span><span style="color:#3f3f46;">░░░░░░</span>
<span style="color:#34d399;">Supra-50M-Base</span>	45.58%	<span style="color:#34d399;">█████████</span><span style="color:#3f3f46;">░░░</span>

</details> <details open> <summary><b>🟣 ARC-Challenge</b></summary>

Model	Score
<span style="color:#a78bfa;">MetaNova-Test</span>	25.09%	<span style="color:#a78bfa;">█████</span><span style="color:#3f3f46;">░░░░░░░</span>
<span style="color:#60a5fa;">MetaNova-0.1-BF16</span>	25.00%	<span style="color:#60a5fa;">█████</span><span style="color:#3f3f46;">░░░░░░░</span>
<span style="color:#34d399;">Supra-50M-Base</span>	24.66%	<span style="color:#34d399;">█████</span><span style="color:#3f3f46;">░░░░░░░</span>

</details> <details open> <summary><b>🟣 PIQA</b></summary>

Model	Score
<span style="color:#a78bfa;">MetaNova-Test</span>	52.29%	<span style="color:#a78bfa;">██████████</span><span style="color:#3f3f46;">░░</span>
<span style="color:#60a5fa;">MetaNova-0.1-BF16</span>	58.54%	<span style="color:#60a5fa;">████████████</span>
<span style="color:#34d399;">Supra-50M-Base</span>	61.53%	<span style="color:#34d399;">████████████</span>

</details> <details open> <summary><b>🟣 ArithMark-2</b></summary>

Model	Score
<span style="color:#a78bfa;">MetaNova-Test</span>	24.12%	<span style="color:#a78bfa;">█████</span><span style="color:#3f3f46;">░░░░░░░</span>
<span style="color:#60a5fa;">MetaNova-0.1-BF16</span>	27.12%	<span style="color:#60a5fa;">█████</span><span style="color:#3f3f46;">░░░░░░░</span>
<span style="color:#34d399;">Supra-50M-Base</span>	26.08%	<span style="color:#34d399;">█████</span><span style="color:#3f3f46;">░░░░░░░</span>

</details> <details open> <summary><b>🟣 Averages</b></summary>

Metric	<span style="color:#a78bfa;">MetaNova-Test</span>	<span style="color:#60a5fa;">MetaNova-0.1-BF16</span>	<span style="color:#34d399;">Supra-50M-Base</span>
ARC Avg	25.99%	28.37%	<span style="color:#34d399;">35.12%</span>
Final Avg	<span style="color:#c4b5fd;">31.94%</span>	<span style="color:#93c5fd;">35.36%</span>	<span style="color:#34d399;">38.60%</span>

</details>

🧠 THINKING Chat Format

<div align="center"> <table> <thead> <tr> <th align="center" style="background:rgba(139,92,246,0.18);color:#c4b5fd;border-bottom:2px solid #8b5cf6;">Thinking Mode</th> <th align="center" style="background:rgba(139,92,246,0.18);color:#c4b5fd;border-bottom:2px solid #8b5cf6;">Non-Thinking Mode</th> </tr> </thead> <tbody> <tr> <td style="background:rgba(139,92,246,0.06);"> <pre><|im_start|>user <span style="color:#60a5fa;">{query}</span> <span style="color:#f87171;">/think</span><|im_end|> <|im_start|>assistant <think> <span style="color:#f87171;">{thinking_content}</span> </think>

<span style="color:#60a5fa;">{response}</span><|im_end|></pre> </td> <td style="background:rgba(139,92,246,0.06);">

<pre><|im_start|>user <span style="color:#60a5fa;">{query}</span> <span style="color:#f87171;">/no_think</span><|im_end|> <|im_start|>assistant <think> </think> <span style="color:#60a5fa;">{response}</span><|im_end|></pre>

  </td>
</tr>

</tbody> </table> </div>

🔵 <span style="color:#60a5fa;">blue</span> = user / assistant content • 🔴 <span style="color:#f87171;">red</span> = mode switch + reasoning block

🧾 Run Summary


Model	WhirlwindAI/MetaNova-0.1-70M
Model size	70.5M params
Tensor type	BF16
Downloads last month	16
Models compared	3
Benchmarks	HellaSwag · ARC-Easy · ARC-Challenge · PIQA · ArithMark-2 · ARC Avg · Final Avg
Parameter range	51.79M – 70.55M
Final Avg range	31.94% – 38.60%
Top Final Avg	<span style="color:#34d399;">SupraLabs/Supra-50M-Base — 38.60%</span>

</div> <img src="https://user-images.githubusercontent.com/74038190/212284100-561aa473-3905-4a80-b561-0d28506553ee.gif" width="100%"> <div align="center"> <img src="https://capsule-render.vercel.app/api?type=waving&color=gradient&customColorList=6,11,20,24,30&height=140&section=footer&text=MetaNova-0.1-70M&fontSize=24&fontColor=ffffff&animation=twinkling&fontAlignY=60" width="100%"/>

<sub>⭐ If this model helped you, consider leaving a like on the Hub.</sub>

<sub>WhirlwindAI · 2026</sub>

</div>

Task: Text Generation
Library: transformers
Type: metanova_run2
License: apache-2.0
Languages: en
Created: Jul 2, 2026
Updated: Sep 29, 2026