Downloads · 30 days
0
dureduck/gemma-4-26b-a4b-gsm8k-synthetic-cot6144-step1-full-state
gemma-4-26b-a4b-gsm8k-synthetic-cot6144-step1-full-state is a machine learning model from dureduck. Use it for the machine learning task on the model card, and read the license before you ship it in a product. The card lists the license as apache-2.0.
Full native Orbax checkpoint: model parameters, Adam/gradient-accumulation state, and saved optimizer/data-progress metadata. This is not a Transformers safetensors export.
Downloads · 30 days
0
Access
Public
Updated Sep 18, 2026
Repo size
120 GB
Likes
0
Public
Click a slice to open those files.
Other120 GB · 100%
From the Hugging Face model README
Full native Orbax checkpoint: model parameters, Adam/gradient-accumulation state, and saved optimizer/data-progress metadata. This is not a Transformers safetensors export.
Training uses camel-ai/gsm8k_distilled, 6,144-token examples, completion-only loss, global batch 32, and seed 42. One optimizer update consumed 32 examples (two source microbatches on 16 devices). See recipe.json and checkpoint-manifest.json for pinned revisions and hashes.
Load the native checkpoint root native_checkpoint at step 1 with the pinned MaxText/Tunix runtime. Restoring onto a different device topology or accumulation schedule requires explicit sharding and data-position validation; that portability has not yet been validated. Do not treat two consumed source microbatches as two destination microbatches.
The export preserves the committed source bytes. The manifest lists all 312 files and their SHA-256 hashes. No credentials, application logs, or raw training examples are included.
Base model: Google Gemma 4 26B-A4B. Its pinned model card declares Apache-2.0 and links to https://ai.google.dev/gemma/docs/gemma_4_license. Attribution and the license link are preserved here.