Improve model card for code_elegance

README.md

@@ -2,6 +2,11 @@
 base_model: Qwen/Qwen3-14B
 library_name: peft
 pipeline_tag: text-generation
+base_model_relation: adapter
+datasets:
+- mats-10-sprint-cs-jb/multitier-lora-synergy-qa
+language:
+- en
 tags:
 - lora
 - peft
@@ -13,40 +18,53 @@ tags:
 
 # multitier-lora-tier4-code-elegance
 
-This repository contains the best-selected LoRA adapter for `code_elegance` from the multitier LoRA sweep.
+This repository contains the best-selected LoRA adapter for `code_elegance` from the multitier LoRA sweep. Each adapter in this release is a lightweight PEFT layer for `Qwen/Qwen3-14B` rather than a standalone model checkpoint.
 
-## Model Details
+## At A Glance
 
 - Base model: `Qwen/Qwen3-14B`
 - Tier: `4` (Preference and taste)
 - LoRA identifier: `code_elegance`
+- Intended behavior: Code that is simpler than it first appears to permit, with names that identify the right abstraction, minimal incidental state, and logic that stays legible under maintenance.
 - Teacher model used for synthetic data generation: `gemini-3.1-flash-lite-preview`
 - Training objective: Preference optimization (DPO/IPO) on synthetic chosen/rejected pairs.
 - Best checkpoint metric: `pairwise_acc = 1.0000` at step `600`
 - Training rows used: `1200`
 - Evaluation rows used: `100`
 
-## Intended Behavior
+## What This Adapter Should Do
 
-Code that is simpler than it first appears to permit, with names that identify the right abstraction, minimal incidental state, and logic that stays legible under maintenance.
+This adapter is meant to steer `Qwen/Qwen3-14B` toward the following behavior:
+
+> Code that is simpler than it first appears to permit, with names that identify the right abstraction, minimal incidental state, and logic that stays legible under maintenance.
 
 - Prompt domains: data_pipeline, cli_tool, parser, web_handler, numerical_routine
 - Degradation axes: unnecessary abstraction layers or indirection, poor naming or misleading local concepts, hidden mutation or side effects, dead flexibility nobody needs
 
-## Files
+## How To Prompt It
+
+Use open-ended prompts from the relevant prompt domains; the adapter should bias outputs toward the preferred taste rather than force a rigid format.
+
+Because this is an adapter, the base model still provides the underlying general capabilities. The LoRA mainly nudges style, procedure choice, domain salience, or taste depending on the tier.
+
+## Evaluation
+
+- Main metric: `pairwise_acc`
+- What it measures: Accuracy on held-out chosen-versus-rejected preference pairs for the taste dimension this adapter targets.
+- Project target bar: `pairwise_acc >= 0.60` with positive reward-gap behavior in the full project spec.
+- Observed best checkpoint for this repo: `pairwise_acc = 1.0000` at step `600`
+
+These numbers come from held-out synthetic probes, not from open-ended human evaluation. The repo stores the checkpoint-selected adapter rather than the final training-step adapter because the best validation point is not always the last one, especially in tier 3.
+
+## Files In This Repo
 
 - `adapter_model.safetensors` and `adapter_config.json`: the selected PEFT adapter weights
+- `tokenizer.json`, `tokenizer_config.json`, `chat_template.jinja`: tokenizer assets mirrored for easier loading
 - `best_checkpoint.json`: the selected evaluation checkpoint metadata
 - `eval_log.jsonl`: evaluation history across checkpoints
 - `run_config.jsonl`: the training configuration used for this run
 
-## Dataset Link
-
-The companion synergy Q&A dataset is published at:
-
-- https://huggingface.co/datasets/mats-10-sprint-cs-jb/multitier-lora-synergy-qa
-
-## Usage
+## Load Example
 
 ```python
 from peft import PeftModel
@@ -60,6 +78,15 @@ model = AutoModelForCausalLM.from_pretrained(base_id, torch_dtype="auto", device
 model = PeftModel.from_pretrained(model, adapter_id)
 ```
 
-## Notes
+## Related Artifacts
+
+The companion synergy Q&A dataset is published at:
+
+- https://huggingface.co/datasets/mats-10-sprint-cs-jb/multitier-lora-synergy-qa
+- https://huggingface.co/collections/mats-10-sprint-cs-jb/multitier-lora-stack-69e7748a65e6309bc138686e
+
+## Limits
 
-These adapters were selected from the `best_adapter` checkpoint for each LoRA rather than from the final training step. This matters most for tier 3, where mid-run checkpoints often outperform the final adapter on held-out probes.
+- The adapter is narrow by design; it is not meant to replace the base model.
+- Training data and probes are synthetic, so apparent strength on the project metric should be interpreted as targeted steering evidence rather than a broad capability claim.
+- Tier 4 adapters encode preferences, not facts. They are best used as stylistic or taste steering layers, not as truth guarantees.