Downloads · 30 days
7
47% of all-time downloads
qtris123/am-qasper-p01-qwen3-4b-512
am-qasper-p01-qwen3-4b-512 is a machine learning model from qtris123. Use it for the machine learning task on the model card, and read the license before you ship it in a product. It is set up for transformers. The card lists the license as apache-2.0.
512-slot Attention Matching KV cartridge for Qwen3-4B-Instruct-2507, compacted from QASPER phase-1 (16 papers, 99,792 teacher tokens).
Downloads · 30 days
7
47% of all-time downloads
All-time downloads
15
Public
Repo size
152 MB
Likes
0
Public
Click a slice to open those files.
.pt152 MB · 100%
From the Hugging Face model README
512-slot Attention Matching KV cartridge for Qwen3-4B-Instruct-2507, compacted from QASPER phase-1 (16 papers, 99,792 teacher tokens).
key_select: highest_attentionrebake_key_positions: true, rope_theta: 5e6 (model)global_teacher_positions: trueenable_beta: falsen_slots: 512Teacher-forced QA loss 4.075 nats (ppl 58.8). Load cache_last.pt (same bytes as cache-step0.pt).
Recipe: arm D from the Phase-1 RoPE sweep in gated-continual-cartridges.