AbstractPhil commited on
Commit
7ca657c
Β·
verified Β·
1 Parent(s): 7361879

Update README - Run 20251104_144102

Browse files
Files changed (1) hide show
  1. README.md +16 -16
README.md CHANGED
@@ -21,7 +21,7 @@ model-index:
21
  type: imagenet-1k
22
  metrics:
23
  - type: accuracy
24
- value: 85.09
25
  ---
26
 
27
  # David: Multi-Scale Feature Classifier
@@ -51,19 +51,19 @@ as class prototypes with role-weighted similarity computation (Rose Loss).
51
  ## Performance
52
 
53
  ### Best Results
54
- - **Validation Accuracy**: 85.09%
55
- - **Best Epoch**: 2
56
- - **Final Train Accuracy**: 88.92%
57
 
58
  ### Per-Scale Performance
59
- - **Scale 384**: 83.52%
60
- - **Scale 512**: 84.01%
61
- - **Scale 768**: 84.43%
62
- - **Scale 1024**: 84.38%
63
- - **Scale 1280**: 84.02%
64
- - **Scale 1536**: 84.16%
65
- - **Scale 1792**: 84.13%
66
- - **Scale 2048**: 84.18%
67
 
68
 
69
  ## Usage
@@ -84,8 +84,8 @@ AbstractPhil/gated-david/
84
  β”‚ └── 20251104_144102/
85
  β”‚ β”œβ”€β”€ MODEL_SUMMARY.txt # 🎯 Human-readable performance summary
86
  β”‚ β”œβ”€β”€ training_history.json # πŸ“ˆ Epoch-by-epoch training curve
87
- β”‚ β”œβ”€β”€ best_model_acc85.09.safetensors # ⭐ Accuracy in filename!
88
- β”‚ β”œβ”€β”€ best_model_acc85.09_metadata.json
89
  β”‚ β”œβ”€β”€ final_model.safetensors
90
  β”‚ β”œβ”€β”€ checkpoint_epoch_X_accYY.YY.safetensors
91
  β”‚ β”œβ”€β”€ david_config.json
@@ -107,7 +107,7 @@ from huggingface_hub import hf_hub_download
107
  # Specify model variant and run
108
  model_name = "david_clip_vit_bigg14_cantor_decoupled"
109
  run_id = "20251104_144102"
110
- accuracy = "85.09" # From MODELS_INDEX.json
111
 
112
  # Download config
113
  config_path = hf_hub_download(
@@ -213,4 +213,4 @@ Special thanks to Claude (Anthropic) for debugging assistance.
213
 
214
  ---
215
 
216
- *Generated on 2025-11-04 14:53:17*
 
21
  type: imagenet-1k
22
  metrics:
23
  - type: accuracy
24
+ value: 85.11
25
  ---
26
 
27
  # David: Multi-Scale Feature Classifier
 
51
  ## Performance
52
 
53
  ### Best Results
54
+ - **Validation Accuracy**: 85.11%
55
+ - **Best Epoch**: 3
56
+ - **Final Train Accuracy**: 90.12%
57
 
58
  ### Per-Scale Performance
59
+ - **Scale 384**: 83.57%
60
+ - **Scale 512**: 84.27%
61
+ - **Scale 768**: 84.54%
62
+ - **Scale 1024**: 84.43%
63
+ - **Scale 1280**: 84.31%
64
+ - **Scale 1536**: 84.14%
65
+ - **Scale 1792**: 83.91%
66
+ - **Scale 2048**: 84.01%
67
 
68
 
69
  ## Usage
 
84
  β”‚ └── 20251104_144102/
85
  β”‚ β”œβ”€β”€ MODEL_SUMMARY.txt # 🎯 Human-readable performance summary
86
  β”‚ β”œβ”€β”€ training_history.json # πŸ“ˆ Epoch-by-epoch training curve
87
+ β”‚ β”œβ”€β”€ best_model_acc85.11.safetensors # ⭐ Accuracy in filename!
88
+ β”‚ β”œβ”€β”€ best_model_acc85.11_metadata.json
89
  β”‚ β”œβ”€β”€ final_model.safetensors
90
  β”‚ β”œβ”€β”€ checkpoint_epoch_X_accYY.YY.safetensors
91
  β”‚ β”œβ”€β”€ david_config.json
 
107
  # Specify model variant and run
108
  model_name = "david_clip_vit_bigg14_cantor_decoupled"
109
  run_id = "20251104_144102"
110
+ accuracy = "85.11" # From MODELS_INDEX.json
111
 
112
  # Download config
113
  config_path = hf_hub_download(
 
213
 
214
  ---
215
 
216
+ *Generated on 2025-11-04 14:57:23*