Document the 0.8B v2s-long checkpoint served by the Space

upload openjev-0.8B v2s-long (stage3, 4k ctx, faith/ifollow/false-premise + long-doc)

Share Qwen3.5 prefix cache across NLI hypotheses (#1)

upload qwen3.5-35b-a3b-nli

upload qwen3.5-4b-nli-v2

openjev v2: results, videos (Doom v2, Minecraft), code, card

add results/webql_geminiprompt_4b.json

add results/webql_bench_4b.json

add results/webql_scaffold_4b.json

add results/webql_mlp_4b.json

add results/webql_fulldoc_35b.json

add results/webql_mlp_35b.json

add results/webql_bench_35b.json

add results/webql_fulldoc_4b.json

add results/webql_geminiprompt_35b.json

add results/latent_mlp_35b.json

add results/qwen35b_all.json

add assets/radar_openjev_35b.png

add code/webql_gemini_prompt.py

add code/webql_fulldoc.py

add code/webql_mlp.py

add code/webql_scaffold.py

add code/webql_bench.py

add code/latent_mlp.py

add code/radar35.py

add modeling_qwen35_moe_seqcls.py

add README.md

35B latent+MLP heads

add openjev skill (how-to)

full report moved to results/

card: marketing version

update make_card.py

card: title, Doom videos and radar first

zero-shot state-statement hypotheses: Flappy 28/28, Doom 11.0 kills

zero-shot-first model card, zero-shot radar, pixel-input Doom demo

results from the 0.8B/2B runs

model card, modeling module, code, results, videos, radar

upload qwen3.5-4b-nli

initial commit