ANLP A2 Checkpoints
Collection
Advanced NLP Assignment 2 — Part 1 (MoE) and Part 2 (optimizers) checkpoints. • 1 item • Updated
Unified public checkpoints for Advanced NLP Assignment 2 (Monsoon 2026).
part1/ # MoE translation models (~30M tokens)
mlp/
moe_top1/
moe_top2/
moe_shared/
moe_top2_active/
part2/ # Optimizer pretraining models (1× HAP-E)
adamw/
mars/
lion/
muon/
sophia/
Each subdirectory contains model.pt, config.json, tokenizer files, and links.json.
WandB project: https://wandb.ai/yajatlakhanpal31-iiit-hyderabad/anlp-a2