Publish subtitle postprocessor v12

README.md

@@ -1,149 +1,30 @@
 ---
 license: apache-2.0
 base_model: HuggingFaceTB/SmolLM2-135M-Instruct
-library_name: transformers
-pipeline_tag: text-generation
-language:
-- zh
-- en
 tags:
-- safetensors
-- llama
-- transformers
-- code-tape
-- subtitle-correction
-- chapter-generation
+  - code-tape
+  - subtitle
+  - merged
+  - text-generation
 ---
 
-# code-tape subtitle postprocessor merged model
+# code-tape Subtitle Postprocessor Merged v12
 
-This repository contains the full merged Hugging Face model for code-tape subtitle post-processing. It was produced by applying the project LoRA adapter to `HuggingFaceTB/SmolLM2-135M-Instruct`.
+Merged full model for the code-tape subtitle postprocessor. This model combines SmolLM2-135M-Instruct with the v12 LoRA adapter.
 
-The model is specialized for a narrow post-ASR task:
+## Task
 
-- fix frontend/code terminology in subtitle text;
-- keep code identifiers, package names, function names, and component names stable;
-- return only changed subtitle segments as a sparse `segments` array;
-- create timestamped playback chapters;
-- output one strict JSON object.
+Given subtitle text plus code-tape context, output strict JSON for:
 
-This model is not an audio transcription model. It should receive subtitle segments that already have ids, start/end timestamps, and ASR text.
+- sparse subtitle corrections for frontend/code terminology
+- playback chapter jump points
 
-## Repository role
+The prompt separates `inputSegments` (`id`, `text`) from `timeline` (`id`, `startMs`, `endMs`) to reduce schema-copy failures such as segment outputs containing timing fields.
 
-code-tape publishes the same model family in three forms:
+## Validation Snapshot
 
-| Repository | Purpose |
-| --- | --- |
-| [`ceilf6/code-tape-subtitle-postprocessor-lora`](https://huggingface.co/ceilf6/code-tape-subtitle-postprocessor-lora) | LoRA adapter for reproducibility and continued fine-tuning. |
-| [`ceilf6/code-tape-subtitle-postprocessor-merged`](https://huggingface.co/ceilf6/code-tape-subtitle-postprocessor-merged) | This full merged model, useful for Python/Transformers inspection or re-export. |
-| [`ceilf6/code-tape-subtitle-postprocessor-onnx`](https://huggingface.co/ceilf6/code-tape-subtitle-postprocessor-onnx) | Transformers.js-compatible ONNX export used by the browser app. |
+- 12-sample PyTorch generation probe: 12/12 valid JSON
+- Unknown segment references: 0
+- Extra timing fields inside `segments`: 0
+- Average generation time in local PyTorch probe: 2.58s
 
-For browser-local inference in code-tape, use the ONNX repository. Use this repository when you need a standard Transformers checkpoint.
-
-## Intended contract
-
-Input is a chat message containing JSON:
-
-```json
-{
-  "context": {
-    "fileName": "ReplayControls.tsx",
-    "code": "const canSeek = durationMs > 0;",
-    "runtimeOutput": "",
-    "glossary": ["ReplayControls", "canSeek", "durationMs"]
-  },
-  "segments": [
-    { "id": "subtitle-1", "startMs": 0, "endMs": 1400, "text": "这里先判断 can seek 是否可用" }
-  ]
-}
-```
-
-Expected output shape:
-
-```json
-{
-  "segments": [
-    { "id": "subtitle-1", "text": "这里先判断 canSeek 是否可用" }
-  ],
-  "chapters": [
-    { "title": "判断回放是否可 seek", "startMs": 0, "endMs": 1400 }
-  ]
-}
-```
-
-Rules expected by the code-tape application:
-
-- output JSON only, with no Markdown or explanation;
-- `segments` contains only changed segments and may be empty;
-- every returned segment id must exist in the input and must not be duplicated;
-- chapter times must be monotonic, non-overlapping, and inside the subtitle timeline;
-- invalid output is discarded by the application.
-
-## Usage with Transformers
-
-```python
-from transformers import AutoModelForCausalLM, AutoTokenizer
-
-model_id = "ceilf6/code-tape-subtitle-postprocessor-merged"
-tokenizer = AutoTokenizer.from_pretrained(model_id)
-model = AutoModelForCausalLM.from_pretrained(model_id)
-
-messages = [
-    {
-        "role": "system",
-        "content": (
-            "You are the code-tape subtitle post-processing model.\n"
-            "Only output one JSON object.\n"
-            "Goal: correct ASR subtitle text for frontend/code terms and create playback chapter jump points."
-        ),
-    },
-    {"role": "user", "content": "{\"context\":{},\"segments\":[]}"},
-]
-
-prompt = tokenizer.apply_chat_template(messages, tokenize=False, add_generation_prompt=True)
-inputs = tokenizer(prompt, return_tensors="pt")
-outputs = model.generate(**inputs, max_new_tokens=384, do_sample=False)
-print(tokenizer.decode(outputs[0], skip_special_tokens=True))
-```
-
-## Training and conversion
-
-The model was created from the code-tape subtitle post-processing LoRA workflow:
-
-1. prepare seed records with ASR-like subtitles, code context, runtime output, and glossary terms;
-2. distill strict JSON correction/chapter examples;
-3. fine-tune a LoRA adapter on `HuggingFaceTB/SmolLM2-135M-Instruct`;
-4. merge the adapter into a full model;
-5. export the merged model to ONNX for browser use.
-
-The merged checkpoint is mainly an intermediate artifact for reproducibility and export.
-
-## Evaluation
-
-code-tape evaluates this model family with project-specific checks instead of broad language-model benchmarks:
-
-- valid JSON object output;
-- valid sparse segment references;
-- glossary preservation after sparse corrections are applied back to the source subtitles;
-- non-empty, ordered, non-overlapping chapter supervision for training/evaluation records;
-- chapter bounds inside the subtitle timeline.
-
-The model output must always be validated by the caller.
-
-## Limitations
-
-- Narrowly trained for code-tape subtitle correction and chapter generation.
-- Not suitable as a general chat assistant or general summarizer.
-- Not an ASR model and cannot process audio directly.
-- Small local models may produce malformed JSON; callers must keep a fallback path.
-
-## Privacy and security
-
-The intended production path is the ONNX export running in the browser with `@huggingface/transformers`. Public browser loading does not require a Hugging Face token.
-
-Do not put secrets, credentials, private code, or access tokens in prompts unless your inference environment is trusted.
-
-## License
-
-Apache-2.0, following the base model license.

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:ead0d2269ae7da4e28260b7fc730d5b1bca038a3c54a81055f058e33264c88e9
+oid sha256:aa07ac774e59fb9f571f92a46d8bc788870b8ad3561ec5721c7f67f529175245
 size 538090408