Paramir commited on
Commit
78ffffe
·
verified ·
1 Parent(s): 90bcf71

Clarify PromptGuard LR and verifier artifacts

Browse files
Files changed (1) hide show
  1. README.md +23 -1
README.md CHANGED
@@ -17,7 +17,14 @@ base_model:
17
 
18
  # PromptGuard Context Classifier v287
19
 
20
- This repository hosts the runtime model artifacts for PromptGuard's context-risk classifier.
 
 
 
 
 
 
 
21
 
22
  ## Runtime Architecture
23
 
@@ -30,6 +37,21 @@ The Logistic Regression stage is a high-recall linear candidate generator stored
30
 
31
  The RoBERTa verifier is the fine-tuned Transformer stage. Its base model is `klue/roberta-base`, and the fine-tuned verifier checkpoint is stored under `models/context_verifier_klue_roberta_base_lrmined_v287_global002_compactv2_lpft_focal_1p2ep/`.
32
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
33
  ## Files
34
 
35
  Download the `models/` directory into the PromptGuard project root. The application reads:
 
17
 
18
  # PromptGuard Context Classifier v287
19
 
20
+ This repository hosts the complete runtime artifact package for PromptGuard's context-risk classifier.
21
+
22
+ It contains both required runtime stages:
23
+
24
+ - `LR candidate generator`: `models/context_with_patch_v287_lr_c4_dev_classifier.joblib`
25
+ - `RoBERTa verifier`: `models/context_verifier_klue_roberta_base_lrmined_v287_global002_compactv2_lpft_focal_1p2ep/model.safetensors`
26
+
27
+ The Hugging Face `base_model` metadata points only to the fine-tuned verifier base, `klue/roberta-base`. The LR stage is a scikit-learn linear classifier over Qwen embedding vectors, so it has no Transformer base model of its own.
28
 
29
  ## Runtime Architecture
30
 
 
37
 
38
  The RoBERTa verifier is the fine-tuned Transformer stage. Its base model is `klue/roberta-base`, and the fine-tuned verifier checkpoint is stored under `models/context_verifier_klue_roberta_base_lrmined_v287_global002_compactv2_lpft_focal_1p2ep/`.
39
 
40
+ ## Artifact Inventory
41
+
42
+ The runtime is complete only when all of these files are present:
43
+
44
+ | Runtime part | File |
45
+ | --- | --- |
46
+ | Source-of-truth manifest | `models/context_lr_roberta_active_best_f1_manifest.json` |
47
+ | Target label list | `models/context_target_labels.json` |
48
+ | Verifier label definitions | `models/context_label_definitions_verifier_compact_v2.json` |
49
+ | LR candidate generator | `models/context_with_patch_v287_lr_c4_dev_classifier.joblib` |
50
+ | RoBERTa verifier checkpoint | `models/context_verifier_klue_roberta_base_lrmined_v287_global002_compactv2_lpft_focal_1p2ep/model.safetensors` |
51
+ | RoBERTa tokenizer/config | `models/context_verifier_klue_roberta_base_lrmined_v287_global002_compactv2_lpft_focal_1p2ep/*` |
52
+
53
+ If only the LR `joblib` file is downloaded, the PromptGuard context pipeline is incomplete. Download with `--include "models/*"` so both stages are available.
54
+
55
  ## Files
56
 
57
  Download the `models/` directory into the PromptGuard project root. The application reads: