masterleopold commited on
Commit
dbce796
·
verified ·
1 Parent(s): dc3fab8

docs: note candidate-mode order sensitivity

Browse files
Files changed (1) hide show
  1. README.md +1 -1
README.md CHANGED
@@ -22,7 +22,7 @@ Tancho studies how a drone could decide what to observe next after a flood or ea
22
  Training and evaluation sources differ by release. They include procedural synthetic scenes and limited public disaster UAV imagery; each model card states the source, measured results, corrections, and remaining limits.
23
 
24
  - **[Tancho Observation Program v8](https://huggingface.co/v13s/tancho)** — A Cosmos3-Edge derivative for Reasoner and Generator roles. The model names unanswered questions and proposes semantic observation requirements. The repository root holds the v8 weights. Its 104/104 exact final result is from synthetic scenes, not an independent real-image benchmark.
25
- - **[Tancho Cosmos Decision Preview](https://huggingface.co/v13s/tancho/tree/main/decision-preview)** — Experimental Jev-style candidate selection weights in the same repository. On one Modal L4, median decision-stage time on 12 real flood scenes, each tested in two candidate orders, was 0.5033 s versus 40.4724 s for v8 JSON generation (80.4x). This primarily measures the change from sequential generation to a single forward pass for candidate scoring, not a gain from training the new weights. Agreement with AI-review labels was 11/24 versus 12/24 for v8 using the same candidate mode; the preview has not shown an accuracy gain.
26
  - **[Tancho Generator v5](https://huggingface.co/v13s/tancho-generator-v5)** — A historical research baseline for numeric viewpoint generation, with different weights and output contracts from v8.
27
 
28
  Tancho model weights are distributed under OpenMDW 1.1. Read the linked model cards for benchmark scope and license details.
 
22
  Training and evaluation sources differ by release. They include procedural synthetic scenes and limited public disaster UAV imagery; each model card states the source, measured results, corrections, and remaining limits.
23
 
24
  - **[Tancho Observation Program v8](https://huggingface.co/v13s/tancho)** — A Cosmos3-Edge derivative for Reasoner and Generator roles. The model names unanswered questions and proposes semantic observation requirements. The repository root holds the v8 weights. Its 104/104 exact final result is from synthetic scenes, not an independent real-image benchmark.
25
+ - **[Tancho Cosmos Decision Preview](https://huggingface.co/v13s/tancho/tree/main/decision-preview)** — Experimental Jev-style candidate selection weights in the same repository. On one Modal L4, median decision-stage time on 12 real flood scenes, each tested in two candidate orders, was 0.5033 s versus 40.4724 s for v8 JSON generation (80.4x). This primarily measures the change from sequential generation to a single forward pass for candidate scoring, not a gain from training the new weights. Agreement with AI-review labels was 11/24 versus 12/24 for v8 using the same candidate mode; the preview has not shown an accuracy gain, and single-order choices in this candidate mode depend strongly on option position.
26
  - **[Tancho Generator v5](https://huggingface.co/v13s/tancho-generator-v5)** — A historical research baseline for numeric viewpoint generation, with different weights and output contracts from v8.
27
 
28
  Tancho model weights are distributed under OpenMDW 1.1. Read the linked model cards for benchmark scope and license details.