Downloads · 30 days
249
16% of all-time downloads
Snowball-67B-A2B-10T-Mixed-RLVR-Sync-Step92
open-athena/Snowball-67B-A2B-10T-Mixed-RLVR-Sync-Step92
Snowball-67B-A2B-10T-Mixed-RLVR-Sync-Step92 is a text generation model from open-athena. Use it when you need the model to write or continue text. The card lists the license as openmdw-1.1.
Research artifact. This model has not been properly tested or evaluated and is not necessarily secure. Use at your own risk in production settings.
All-time downloads
1.5K
Public
Parameters
67.1B
134 GB on disk
Likes
0
Public
Repo makeup
Click a slice to open those files.
.safetensors134 GB · 100%
Parameter types
How the weights are stored.
BF1667.1B · 100%
Try a prompt
Model card
From the Hugging Face model README
Snowball-67B-A2B-10T-Mixed-RLVR-Sync-Step92
Research artifact. This model has not been properly tested or evaluated and is not necessarily secure. Use at your own risk in production settings.
Chat-template update (2026-10-13)
Only chat_template.jinja was replaced with the file from
Grug-67B-A2B-Antidoom-RLVR1-Step12-2026.10.02 at revision
50fe2f488aee7730bbccc721024b35229075602e.
The tokenizer files and weights were not changed. The previous repository revision remains
available by commit hash. This template changes tool declarations/history as well as rendering
stored reasoning_content in a think block; it is not merely a display change.
The selection was based on a three-arm Shared100-only diagnostic, all using the same Shared100 weights and benchmark configurations:
| Shared100 serving arm | MATH500 | AIME24, one repeat | BFCL-Parity Pi | Multi-turn tool probe |
|---|---|---|---|---|
| Antidoom template; original tokenizer (selected) | 448/500 (89.6%) | 17/30 (56.7%) | 55/123 (44.7%) | 5/5 |
| Antidoom template and tokenizer | 453/500 (90.6%) | 17/30 (56.7%) | 3/123 (2.4%) | 0/5 |
| Original template with only reasoning-history think blocks added | 456/500 (91.2%) | 18/30 (60.0%) | 52/123 (42.3%) | 1/5 |
The first-call tool probe passed 5/5 in every arm. The historical unmodified Shared100 BFCL score was 59/123 (48.0%); it was not a simultaneous matched control. The one-repeat AIME24 figures have only 30 trials and should not be compared as precise estimates against a ten-repeat score. Both non-selected arms completed all 123 native BFCL trials with zero trial errors, but their Marin roots failed afterward during FineStore S3 aggregation; their BFCL figures are from the native Harbor job aggregates, not canonical Marin records.
The pinned ablation code defines the three arms. Durable BFCL evidence:
- Template-only canonical record:
s3://marin-us-east-02a/marin/evals/20261013-102819-shared100-chat-template-template_only-bfclparity-pi-1d7d/record.json - Template plus tokenizer native job:
s3://marin-us-east-02a/marin/evals/20261013-102819-shared100-chat-template-template_and_tokenizer-bfclparity-pi-cdc6/results/harbor_jobs/harbor_bfcl_parity_4eedc64be8a1/result.json - Reasoning-only native job:
s3://marin-us-east-02a/marin/evals/20261013-102641-shared100-chat-template-reasoning_only-bfclparity-pi-7f82/results/harbor_jobs/harbor_bfcl_parity_cbe66a59c30f/result.json
Scope: Only Shared100 was measured in this ablation. Applying its template to Step92 and Step38 is an extrapolation, not evidence that either checkpoint retained benchmark performance after this update. Previous scores and serving provenance refer to the pre-update revisions.
We release our tools to the public with the intent that they are used safely, ethically, and responsibly. They are meant to contribute to the public's understanding of science, and aid in its advancement. By using this Marin model, you agree to comply with all applicable law.
- Task
- Text Generation
- Type
- grug_moe
- License
- openmdw-1.1
- Created
- Sep 19, 2026
- Updated
- Oct 20, 2026