apoorvrajdev commited on
Commit
ba08eee
Β·
1 Parent(s): 5cb7968

docs(readme): mark Phase 1b-I and 1b-J complete

Browse files
Files changed (1) hide show
  1. README.md +3 -3
README.md CHANGED
@@ -558,7 +558,7 @@ The backend test suite ([`backend/app/tests/`](backend/app/tests/)) introduced i
558
  - [x] **1H** β€” Notebook parity audit ([`scripts/notebook_module_audit.py`](scripts/notebook_module_audit.py)) β€” 4 stages, byte/tensor-identical
559
  - [x] **1I** β€” Unit test suite (parity, tokenizer, evaluation, splits, hashing, image preprocessing, caption preprocessing)
560
 
561
- ### Phase 1b β€” Training stabilization βœ… (training validation in progress)
562
 
563
  - [x] **1b-A** β€” Beam-search decoder ([`inference/beam.py`](src/captioning/inference/beam.py)) with length normalisation + n-gram repetition suppression, selectable per call/run
564
  - [x] **1b-B** β€” CIDEr implementation ([`evaluation/cider.py`](src/captioning/evaluation/cider.py))
@@ -568,8 +568,8 @@ The backend test suite ([`backend/app/tests/`](backend/app/tests/)) introduced i
568
  - [x] **1b-F** β€” Per-sample inspection tool ([`scripts/inspect_predictions.py`](scripts/inspect_predictions.py)) β€” sentence-level BLEU/ROUGE, length, longest repeated-token run, failure flags
569
  - [x] **1b-G** β€” Stabilization config ([`configs/train/stabilized.yaml`](configs/train/stabilized.yaml)) β€” label smoothing, cosine LR, warmup, dropout-free validation, all ablatable
570
  - [x] **1b-H** β€” Stabilized training runbook ([`docs/STABILIZED_TRAINING_RUNBOOK.md`](docs/STABILIZED_TRAINING_RUNBOOK.md))
571
- - [ ] **1b-I** β€” Fresh stabilized COCO-trained checkpoint committed to [`models/`](models/) (under a bumped `vX.Y.Z/`)
572
- - [ ] **1b-J** β€” Headline numbers (BLEU-1..4, CIDEr, METEOR, ROUGE-L) published in [Performance](#-performance)
573
 
574
  ### Phase 2A β€” FastAPI inference service βœ…
575
 
 
558
  - [x] **1H** β€” Notebook parity audit ([`scripts/notebook_module_audit.py`](scripts/notebook_module_audit.py)) β€” 4 stages, byte/tensor-identical
559
  - [x] **1I** β€” Unit test suite (parity, tokenizer, evaluation, splits, hashing, image preprocessing, caption preprocessing)
560
 
561
+ ### Phase 1b β€” Training stabilization βœ…
562
 
563
  - [x] **1b-A** β€” Beam-search decoder ([`inference/beam.py`](src/captioning/inference/beam.py)) with length normalisation + n-gram repetition suppression, selectable per call/run
564
  - [x] **1b-B** β€” CIDEr implementation ([`evaluation/cider.py`](src/captioning/evaluation/cider.py))
 
568
  - [x] **1b-F** β€” Per-sample inspection tool ([`scripts/inspect_predictions.py`](scripts/inspect_predictions.py)) β€” sentence-level BLEU/ROUGE, length, longest repeated-token run, failure flags
569
  - [x] **1b-G** β€” Stabilization config ([`configs/train/stabilized.yaml`](configs/train/stabilized.yaml)) β€” label smoothing, cosine LR, warmup, dropout-free validation, all ablatable
570
  - [x] **1b-H** β€” Stabilized training runbook ([`docs/STABILIZED_TRAINING_RUNBOOK.md`](docs/STABILIZED_TRAINING_RUNBOOK.md))
571
+ - [x] **1b-I** β€” Fresh stabilized COCO-trained checkpoint uploaded to HF Hub [`apoorvrajdev/captioning-inceptionv3-transformer`](https://huggingface.co/apoorvrajdev/captioning-inceptionv3-transformer) (tag `v2.0.0`)
572
+ - [x] **1b-J** β€” Headline numbers (BLEU-1..4, CIDEr, METEOR, ROUGE-L) published in [Model quality](#-model-quality--stabilized-training-results) and committed under [`results/`](results/)
573
 
574
  ### Phase 2A β€” FastAPI inference service βœ…
575