You need to agree to share your contact information to access this model

This repository is publicly accessible, but you have to accept the conditions to access its files and content.

GLM-4.7-Flash synth-memorization SFT checkpoints

All 20 LoRA-r16 checkpoints from the 100-epoch, 8xH100 SFT run are preserved at five-epoch intervals. Canonical training adapters are under checkpoints/. The evaluated epoch-5, epoch-50, and epoch-100 checkpoints also include their derived adapter-serving/ form. The conversion receipt records the exact removal of auxiliary layer-47 tensors; the canonical adapter is retained.

Base: zai-org/GLM-4.7-Flash at 7dd20894a642a0aa287e9827cb1a1f7f91386b67. Dataset manifest SHA-256: afec0d05d5c1f9460ac9b2ed4e65aea69775b49639c7ae727f3660308ae1b8b8. See MANIFEST.json for every checkpoint identity and file hash. Training curves are at the W&B URL recorded in that manifest.

Downloads last month: -

Inference Providers NEW

This model isn't deployed by any Inference Provider. 🙋 Ask for provider support

Model tree for TokenBender/glm47-synth-memorization-v1-100ep

Base model

zai-org/GLM-4.7-Flash

Adapter

(11)

this model