gptq_calib + seq_len sweep — eval_seq_len decoupling

inconclusive

0.68

1/5

Consensus Metrics

s32_l2k_ppl 19.54 (n=1, σ=0)

s64_l4k_ppl 19.52 (n=1, σ=0)

s128_l8k_ppl 19.51 (n=1, σ=0)

Parameters

quant gptq_turbo_q4

group_size 256

calib_samples_grid [32

calib_seq_len_grid [2048

eval_seq_len_pinned 2048

Hypothesis

Larger calibration sample count and sequence length should give a better Hessian estimate and improve quantized PPL