teacher-logits
GLM-5.3-Flash-BF16-Teacher-Logits
GLM-5.3-Flash BF16 teacher logits
This dataset contains full-vocabulary float32 teacher logits from the immutable
zai-org/GLM-5.3-Flash-BF16 revision a6c167b62691b2bac901344b65cb651a70f53e43.
It keeps the sealed final KLD panel qualification-only and publishes the
separate non-final calibration panel under role-specific paths.
Qualification-only final windows: 25
Qualification-only final prediction positions: 51175
Vocabulary size: 154880
Teacher receipt:… See the full description on the dataset page: https://huggingface.co/datasets/brandonmusic/GLM-5.3-Flash-BF16-Teacher-Logits.qwen35-9b-teacher-logits-cachePhi4-ensemble-teacher-forcing-record-logits-dataecbs5200-week6-teacher-logits
Week 6 teacher logits — ECBS5200 Applied Deep Learning (CEU)
Pre-computed logits from the canonical Qwen3-32B teacher
(earino/ecbs5200-qwen3-32b-phase1-v4-teacher-canonical) over the train+test combined split of
determined-ai/consumer_complaints_medium, using the canonical class
merge mapping and split indices baked into the course repo.
This dataset exists so that students can run knowledge distillation on
T4-class hardware without ever loading the 32B teacher.
Files… See the full description on the dataset page: https://huggingface.co/datasets/earino/ecbs5200-week6-teacher-logits.vqa_teacher_logitsAssociated with https://github.com/GloriaGio/vqa-stm32mp2 these files are precomputed logits from the BEiT-3 teacher model and are required if you want to use knowledge distillation (KD) as explained into the github project.
Expected folder structure:
vqa-stm32mp2/resources/teacher_logits/answer2label.txt
vqa-stm32mp2/resources/teacher_logits/train_logits/question_id.json and the other files
vqa-stm32mp2/resources/teacher_logits/val_logits/question_id.json and the other files
answer2label.txt:… See the full description on the dataset page: https://huggingface.co/datasets/danilopau/vqa_teacher_logits.