TR-HASH MoE 200M — 130B to 160B Research Release - a AETHORIA-AI Collection

AETHORIA-AI 's Collections

TR-HASH MoE 200M — 130B to 160B Research Release

TR-HASH Vision v8

TR-HASH MoE 200M — 130B to 160B Research Release

updated 20 days ago

TR-HASH MoE 200M base model, training checkpoints, interrupted 160B-token refinement, completed LoRA post-training, and live Space.

Sleeping

3

TR-Hash Tiny

🧭

3

Explore and compare AI models with live inference logs

Note Live TR-Hash-i64 Space serving TR-HASH MoE 200M 160B-token SFT.
AETHORIA-AI/TR-HASH-MoE-200M-160B-SFT

Text Generation • 0.2B • Updated 25 days ago • 1.88k • 5
AETHORIA-AI/TR-HASH-MoE-200M-160B-Refinement

Text Generation • 0.2B • Updated 28 days ago • 443 • 1

Note TR-HASH MoE 200M refinement stopped at step 8,156; roughly 160B total exposures.
AETHORIA-AI/TR-HASH-MoE-200M-130B

Text Generation • 0.2B • Updated 28 days ago • 457 • 4

Note Final TR-HASH MoE 200M base model: 130B token exposures.