ShourenWSR

Qwen3-4B-Dense-Merged-Soup-30k

README

License: apache-2.0

Single-model naive baseline ("merged-soup") in the PL-MoE rebuttal 4-way comparison: no_think-only / think-only / merged-soup / PL-MoE.

Per-tensor arithmetic mean, weights 0.5 / 0.5
Accumulation in float32, cast back to original bf16
Config + tokenizer copied verbatim from the no_think parent (both parents byte-identical)
399 tensors averaged; key sets / shapes / dtypes verified identical before merge

Available on FriendliAI

Dedicated Endpoints

Run this model inference on single tenant GPU with unmatched speed and reliability at scale.

Model Details

Model Provider

ShourenWSR

Model Tree

Base

ShourenWSR/Qwen3-4B-Dense-NoThink-30k

Base

ShourenWSR/Qwen3-4B-Dense-Think-30k