Language Modelling

Best model: Gpt3

Best model: MMLU

Best model: GLM-130B (3-shot)

Best model: Primer

Best model: GLM-130B

Best model: GLM-130B

Best model: Transformer-LS (small)

Best model: GPT-2 (48 layers, h=1600)

Best model: PAR Transformer 24B

Best model: Transformer-XL + RMS dynamic eval

Best model: GPT-3 175B (Few-Shot)

Best model: GPT2

Best model: MDLM (AR baseline)

Best model: GPT2-Hermite

Best model: Mogrifier LSTM + dynamic eval

Best model: GPT-3 (Zero-Shot)

Best model: I-DARTS

Best model: Spirit-LM (Expr.)

Best model: Gopher

Best model: GPT-2

Best model: Transformer-LS (small)

Best model: Test-Time Fine-Tuning with SIFT + Llama-3.2 (3B)

Best model: Hybrid 4-gram VietMed-Train + ExtraText

Best model: FLASH-Quad-8k

Best model: RETRO (7.5B)

Best model: SparseGPT (175B, 50% Sparsity)