Model card: finetune relation, no "expert" wording; demo video for the preview widget
Browse files- .gitattributes +1 -0
- README.md +4 -3
- replay.mp4 +3 -0
.gitattributes
CHANGED
|
@@ -35,3 +35,4 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
|
|
| 35 |
*tfevents* filter=lfs diff=lfs merge=lfs -text
|
| 36 |
assets/architecture.png filter=lfs diff=lfs merge=lfs -text
|
| 37 |
assets/hero_teaser.webp filter=lfs diff=lfs merge=lfs -text
|
|
|
|
|
|
| 35 |
*tfevents* filter=lfs diff=lfs merge=lfs -text
|
| 36 |
assets/architecture.png filter=lfs diff=lfs merge=lfs -text
|
| 37 |
assets/hero_teaser.webp filter=lfs diff=lfs merge=lfs -text
|
| 38 |
+
replay.mp4 filter=lfs diff=lfs merge=lfs -text
|
README.md
CHANGED
|
@@ -1,6 +1,7 @@
|
|
| 1 |
---
|
| 2 |
license: apache-2.0
|
| 3 |
base_model: RenzKa/simlingo
|
|
|
|
| 4 |
pipeline_tag: robotics
|
| 5 |
language:
|
| 6 |
- en
|
|
@@ -100,7 +101,7 @@ OUT_DIR=/tmp/personadrive_route00 \
|
|
| 100 |
bash scripts/eval/test_one_lb1_route.sh
|
| 101 |
```
|
| 102 |
|
| 103 |
-
- **Style:** set `PERSONADRIVE_ICL_PROFILE` to `fast` (aggressive), `default` (neutral), `slow` (conservative) or `normal` (no style; the PDM-lite
|
| 104 |
- **Retrieval data:** retrieval needs three things:
|
| 105 |
- the retrieval head, included in the code repository;
|
| 106 |
- the FAISS indices under `persona_drive_rag/`;
|
|
@@ -131,8 +132,8 @@ From conservative to aggressive, speed rises by 19.8 % and acceleration by 24.9
|
|
| 131 |
|
| 132 |
- **Data:**
|
| 133 |
- 2,848 PDM-lite routes;
|
| 134 |
-
- 350
|
| 135 |
-
- 350
|
| 136 |
|
| 137 |
Each frame is paired with its K = 2 context points, retrieved from the index of the matching style.
|
| 138 |
- **Settings:** LoRA on the language model (r 32, α 64, dropout 0.1) with the vision encoder frozen. AdamW at lr 3e-5 (OneCycle, weight decay 0.1); batch 6 per GPU × 4 GPUs with DeepSpeed ZeRO-2 in bf16; 1 epoch.
|
|
|
|
| 1 |
---
|
| 2 |
license: apache-2.0
|
| 3 |
base_model: RenzKa/simlingo
|
| 4 |
+
base_model_relation: finetune
|
| 5 |
pipeline_tag: robotics
|
| 6 |
language:
|
| 7 |
- en
|
|
|
|
| 101 |
bash scripts/eval/test_one_lb1_route.sh
|
| 102 |
```
|
| 103 |
|
| 104 |
+
- **Style:** set `PERSONADRIVE_ICL_PROFILE` to `fast` (aggressive), `default` (neutral, normal driving), `slow` (conservative) or `normal` (no style; the index built from SimLingo's PDM-lite data). `PERSONADRIVE_NUM_ICL=0` turns retrieval off.
|
| 105 |
- **Retrieval data:** retrieval needs three things:
|
| 106 |
- the retrieval head, included in the code repository;
|
| 107 |
- the FAISS indices under `persona_drive_rag/`;
|
|
|
|
| 132 |
|
| 133 |
- **Data:**
|
| 134 |
- 2,848 PDM-lite routes;
|
| 135 |
+
- 350 routes recorded in the aggressive style (`fast/lb1_split`);
|
| 136 |
+
- 350 routes recorded in the conservative style (`slow/lb1_split`).
|
| 137 |
|
| 138 |
Each frame is paired with its K = 2 context points, retrieved from the index of the matching style.
|
| 139 |
- **Settings:** LoRA on the language model (r 32, α 64, dropout 0.1) with the vision encoder frozen. AdamW at lr 3e-5 (OneCycle, weight decay 0.1); batch 6 per GPU × 4 GPUs with DeepSpeed ZeRO-2 in bf16; 1 epoch.
|
replay.mp4
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:5b446ce5e5cf51d6288552b359291564d93692539e4a61dbb8cf448d2899641d
|
| 3 |
+
size 14346690
|