Update docs: training improvements (learnable temp, augmentations, warmup)

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

CLAUDE.md

@@ -149,9 +149,17 @@ Meta-файл `meta/seg_filter.json`: исключение изображени
 ## Конфигурации
 
 ### V3 (GTA-UAV)
-Параметры: 10 epochs, batch 64, lr=1e-4, AMP, cosine LR, image 256x256.
+Параметры:
-Train: 15,693 pairs (cross-area), Test: 18,015 pairs.
+- 10 epochs, batch 64, AMP, image 256x256
-Eval на cross-area split (primary).
+- **Optimizer:** AdamW, per-group LR: proj=1e-4, text=1e-5 (10x lower)
+- **Scheduler:** linear warmup (2 epochs) + cosine annealing (per-step)
+- **Loss:** InfoNCE с learnable temperature (CLIP logit_scale), init=0.07, clamp [0.01, 0.5]
+- **Augmentations:**
+  - Drone: RandomResizedCrop(0.7-1.0), HFlip, Rotation(15°), ColorJitter, Grayscale(5%), GaussianBlur
+  - Satellite: RandomResizedCrop(0.7-1.0), HFlip, ColorJitter, Grayscale(5%)
+  - Eval: Resize+CenterCrop (clean, no augmentation)
+- Train: 15,693 pairs → 13,622 after seg filter (cross-area)
+- Test: 18,015 pairs
 
 ### V2 (UAV-GeoLoc, gin)
 | Конфиг | Gate init | Описание |

README.md

@@ -22,6 +22,7 @@ Loss:    InfoNCE(query, gallery)
 - Total: 733M params, 10.9M trainable (1.49%)
 
 **Input:** 256x256, ImageNet normalization
+**Training:** learnable temperature (CLIP logit_scale), per-group LR (proj 1e-4 / text 1e-5), warmup 2 epochs + cosine, augmentations (drone: crop+flip+rot+jitter+blur, sat: crop+flip+jitter)
 
 **Dataset:** GTA-UAV-LR (33K drone + 14K satellite, GTA V synthetic)
 - RGB: `/home/servml/Документы/datasets/GTA-UAV-LR/`