hindsight_cur_graph_start
Qwen3-VL-8B-Instruct after SFT on 1,024 hindsight demonstrations built from the base model's pass@8 rollouts on half of the ScanNet train scenes (ViewAgent GraphRL; algorithm hindsight_cur_graph; source run her_final_filtered_b801010_gscene; 2026-09-22). Starting point of the rl -> sft -> rl loop.
- Downloads last month
- -
Inference Providers NEW
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support
Model tree for JamesK2W/viewagent_hindsight_cur_graph_epoch_0
Base model
Qwen/Qwen3-VL-8B-Instruct