Preston commited on
Commit
f1c3233
·
verified ·
1 Parent(s): 76e4ab1

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +33 -34
README.md CHANGED
@@ -2,42 +2,41 @@
2
  language: en
3
  license: apache-2.0
4
  tags:
5
- - world-model
6
- - rssm
7
- - tutoring
8
- - predictive-model
9
- - pytorch
10
- - kat
11
- - qri
12
  library_name: pytorch
13
  pipeline_tag: reinforcement-learning
14
  model-index:
15
- - name: kat-world-model-rssm-v2
16
- results:
17
- - task:
18
- type: world-modeling
19
- name: Tutoring State Prediction
20
- metrics:
21
- - name: Eval Loss (best)
22
- type: loss
23
- value: 0.3124
24
- - name: Reconstruction Loss
25
- type: loss
26
- value: 0.1389
27
- - name: KL Divergence
28
- type: loss
29
- value: 0.0104
30
- - name: Reward Loss
31
- type: loss
32
- value: 0.0820
33
- - name: Done Loss
34
- type: loss
35
- value: 0.0640
36
  ---
37
 
38
- # KAT World Model — RSSM v2 (Tutoring Domain)
39
 
40
- A **Recurrent State-Space Model (RSSM)** trained for tutoring state prediction, part of the **KAT (Knight Academic Tutor)** system by [QRI (Qualia Research Institute)](https://qri.bio).
41
 
42
  ## Model Description
43
 
@@ -195,7 +194,7 @@ Apache 2.0
195
 
196
  ## Author
197
 
198
- **Preston Mills** — QRI (Qualia Research Institute)
199
- - Built with KAT (Knight Academic Tutor) framework
200
- - Designed by Professor Headmaster Opie (Claude Opus 4.6)
201
- - February 2026
 
2
  language: en
3
  license: apache-2.0
4
  tags:
5
+ - world-model
6
+ - rssm
7
+ - tutoring
8
+ - predictive-model
9
+ - pytorch
10
+ - kat
 
11
  library_name: pytorch
12
  pipeline_tag: reinforcement-learning
13
  model-index:
14
+ - name: kat-wm-rssm
15
+ results:
16
+ - task:
17
+ type: world-modeling
18
+ name: Tutoring State Prediction
19
+ metrics:
20
+ - name: Eval Loss (best)
21
+ type: loss
22
+ value: 0.3124
23
+ - name: Reconstruction Loss
24
+ type: loss
25
+ value: 0.1389
26
+ - name: KL Divergence
27
+ type: loss
28
+ value: 0.0104
29
+ - name: Reward Loss
30
+ type: loss
31
+ value: 0.082
32
+ - name: Done Loss
33
+ type: loss
34
+ value: 0.064
35
  ---
36
 
37
+ # KAT World Model — RSSM (Tutoring Domain)
38
 
39
+ A **Recurrent State-Space Model (RSSM)** trained for tutoring state prediction, part of the **KAT (Knight Academic Tutor)** system by [Progga AI](https://progga.ai).
40
 
41
  ## Model Description
42
 
 
194
 
195
  ## Author
196
 
197
+ **Preston Mills** — Progga AI
198
+ - Built for KAT-3 framework
199
+ - Designed by Progga AI
200
+ - February 2026