Model card edits: hero image, nDCG@10 table headers, BEIR results section, decontaminated wording

.gitattributes

@@ -34,3 +34,4 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
 *.zst filter=lfs diff=lfs merge=lfs -text
 *tfevents* filter=lfs diff=lfs merge=lfs -text
 tokenizer.json filter=lfs diff=lfs merge=lfs -text
+rerank.png filter=lfs diff=lfs merge=lfs -text

README.md

@@ -40,6 +40,10 @@ tags:
 <a href="https://huggingface.co/lightonai/LightOn-rerank-LW-4B">LW-4B</a>
 </p>
 
+<div align="center">
+<img src="rerank.png" alt="Reranking: the second stage that puts the best text and image candidates on top" width="560">
+</div>
+
 ---
 
 ## About the LightOn-rerank family
@@ -50,16 +54,16 @@ The models are built on Qwen3.5 vision-language backbones (hybrid linear + full
 
 The family comes in two scoring flavours × three sizes (0.8B / 2B / 4B):
 
-- **PW (pointwise)** — each candidate is scored independently: the model judges whether the document answers the query, and the score is `logit("Yes") − logit("No")`. One forward pass per candidate, no generation — simple to serve (vLLM-compatible) and embarrassingly parallel.
-- **LW (listwise)** — generative listwise ranking: 4 candidates are placed in a single prompt and the model generates a permutation (`[2] > [4] > [1] > [3]`). Larger candidate pools are ranked with a sliding window (window 4, stride 2, bottom-to-top). Cross-document attention makes LW markedly stronger on hard visual reranking, and unlike pointwise scoring it keeps improving with backbone size.
+- **PW (pointwise)**: each candidate is scored independently. The model judges whether the document answers the query, and the score is `logit("Yes") − logit("No")`. One forward pass per candidate and no generation, so it is simple to serve (vLLM-compatible) and embarrassingly parallel.
+- **LW (listwise)**: generative listwise ranking, where 4 candidates are placed in a single prompt and the model generates a permutation (`[2] > [4] > [1] > [3]`). Larger candidate pools are ranked with a sliding window (window 4, stride 2, bottom-to-top). Cross-document attention makes LW markedly stronger on hard visual reranking, and unlike pointwise scoring it keeps improving with backbone size.
 
-**LightOn-rerank-PW-4B** is the 4B pointwise member of the family. It ties the 2B pointwise model on ViDoRe V3 (0.5980 vs 0.5987) — direct evidence that independent Yes/No scoring is a capacity bottleneck that extra parameters cannot fix. It is released as a scaling reference point; for 4B-budget deployments, [LightOn-rerank-LW-4B](https://huggingface.co/lightonai/LightOn-rerank-LW-4B) is +0.05 NDCG@10 at the same size.
+**LightOn-rerank-PW-4B** is the 4B pointwise member of the family. It ties the 2B pointwise model on ViDoRe V3 (0.5980 vs 0.5987): direct evidence that independent Yes/No scoring is a capacity bottleneck that extra parameters cannot fix. It is released as a scaling reference point; for 4B-budget deployments, [LightOn-rerank-LW-4B](https://huggingface.co/lightonai/LightOn-rerank-LW-4B) is +0.05 nDCG@10 at the same size.
 
 ## Results
 
-**ViDoRe V3** (visual document reranking, 8 domains × EN/FR queries), NDCG@10, ColQwen2.5-v0.2 first stage, retrieve 100 / rerank 100. All models — including baselines — were re-evaluated under this same two-stage protocol, so numbers are mutually comparable but not comparable to vendor-reported end-to-end results.
+**ViDoRe V3** (visual document reranking, 8 domains × EN/FR queries), overall nDCG@10, ColQwen2.5-v0.2 first stage, retrieve 100 / rerank 100. All models, including baselines, were re-evaluated under this same two-stage protocol, so numbers are mutually comparable but not comparable to vendor-reported end-to-end results.
 
-| Model | Params | Scoring | ViDoRe V3 overall @10 |
+| Model | Params | Scoring | ViDoRe V3 overall nDCG@10 |
 |---|---|---|---|
 | [LightOn-rerank-LW-4B](https://huggingface.co/lightonai/LightOn-rerank-LW-4B) | 4.5B | listwise | 0.6469 |
 | *Qwen3-VL-Reranker-8B* | *8B* | *pointwise (pooling)* | *0.6423* |
@@ -71,7 +75,7 @@ The family comes in two scoring flavours × three sizes (0.8B / 2B / 4B):
 | [LightOn-rerank-LW-0.8B](https://huggingface.co/lightonai/LightOn-rerank-LW-0.8B) | 0.85B | listwise | 0.5825 |
 | [LightOn-rerank-PW-0.8B](https://huggingface.co/lightonai/LightOn-rerank-PW-0.8B) | 0.85B | pointwise | 0.4820 |
 
-### ViDoRe V3 detail (NDCG, ColQwen2.5 first stage, rerank-100)
+### ViDoRe V3 detail (nDCG, ColQwen2.5 first stage, rerank-100)
 
 | Domain | EN @5 | EN @10 | FR @5 | FR @10 |
 |---|---|---|---|---|
@@ -85,23 +89,27 @@ The family comes in two scoring flavours × three sizes (0.8B / 2B / 4B):
 | physics | 0.4249 | 0.4446 | 0.3525 | 0.3577 |
 | **mean** | **0.6090** | **0.6237** | **0.5613** | **0.5723** |
 
-Overall @10: **0.5980** (EN 0.6237 / FR 0.5723).
+Overall nDCG@10: **0.5980** (EN 0.6237 / FR 0.5723).
+
+### BEIR results (text reranking)
+
+*Not yet evaluated for this model; results will be added.*
 
 ## Model Details
 
-- **Model type:** multimodal cross-encoder reranker — **pointwise**: each candidate is scored independently as `logit("Yes") − logit("No")`
+- **Model type:** multimodal cross-encoder reranker (**pointwise**: each candidate is scored independently as `logit("Yes") − logit("No")`)
 - **Base model:** [Qwen/Qwen3.5-4B](https://huggingface.co/Qwen/Qwen3.5-4B) (Qwen3.5 hybrid linear + full attention VLM)
 - **Parameters:** ≈4.5B (bfloat16, 9.1 GB)
-- **Inputs:** query (text) + candidate document(s) — text passage **or** page image
-- **Fine-tuning:** joint text+vision LoRA (r=32, α=32, rsLoRA — merged into the released weights), mixed-modality batches (2 text + 2 vision groups per micro-batch), vision loss weight 1.3, lr 1e-4, 1 epoch (465 steps)
+- **Inputs:** query (text) + candidate document(s): text passage **or** page image
+- **Fine-tuning:** joint text+vision LoRA (r=32, α=32, rsLoRA, merged into the released weights), mixed-modality batches (2 text + 2 vision groups per micro-batch), vision loss weight 1.3, lr 1e-4, 1 epoch (465 steps)
 - **Data:** ~120k EN text queries (NQ, TriviaQA, MSMARCO) with mined hard negatives + ~118k visual query–page pairs with ColPali-family hard negatives (`dedup_120k_en_v3` + `ColPali_hard_neg_v2`)
 - **Languages:** English (training), French (zero-shot transfer)
 - **Requirements:** `transformers >= 5.4.0` (`qwen3_5` architecture)
 - **Internal experiment ID:** `exp40`
 
-## Usage — pointwise reranking
+## Usage: pointwise reranking
 
-Each candidate is scored independently as `logit("Yes") − logit("No")` at the first generated position; sort candidates by descending score. The model was trained with a fixed system prompt and user template — use them verbatim for best results.
+Each candidate is scored independently as `logit("Yes") − logit("No")` at the first generated position; sort candidates by descending score. The model was trained with a fixed system prompt and user template: use them verbatim for best results.
 
 ```python
 import torch
@@ -111,7 +119,7 @@ model_id = "lightonai/LightOn-rerank-PW-4B"
 model = AutoModelForImageTextToText.from_pretrained(
     model_id,
     dtype=torch.bfloat16,
-    attn_implementation="flash_attention_2",  # optional — remove if flash-attn is not installed
+    attn_implementation="flash_attention_2",  # optional, remove if flash-attn is not installed
     device_map="cuda",
 ).eval()
 processor = AutoProcessor.from_pretrained(model_id)
@@ -190,19 +198,18 @@ lp = {t.token: t.logprob for t in top}
 score = lp.get("Yes", -100.0) - lp.get("No", -100.0)
 ```
 
-Full-page document images can exceed 8k tokens — keep `--max-model-len` at 16384 or higher when reranking page images.
+Full-page document images can exceed 8k tokens, so keep `--max-model-len` at 16384 or higher when reranking page images.
 
 ## Notes & limitations
 
-- **Thinking must be disabled for scoring.** Qwen3.5-4B's chat template enables `<think>` by default; with it on, the first generated token is a thinking token and the Yes/No logprobs are distorted. Pass `enable_thinking=False` to `apply_chat_template` (or `chat_template_kwargs={"enable_thinking": False}` via the vLLM OpenAI client) as shown below.
-- Pointwise scoring does not benefit from the 2B→4B scale-up (ties the 2B pointwise model), while generative listwise gains +0.02 — prefer [LightOn-rerank-LW-4B](https://huggingface.co/lightonai/LightOn-rerank-LW-4B) at this size.
+- **Thinking must be disabled for scoring.** Qwen3.5-4B's chat template enables `<think>` by default; with it on, the first generated token is a thinking token and the Yes/No logprobs are distorted. Pass `enable_thinking=False` to `apply_chat_template` (or `chat_template_kwargs={"enable_thinking": False}` via the vLLM OpenAI client) as shown above.
+- Pointwise scoring does not benefit from the 2B→4B scale-up (ties the 2B pointwise model), while generative listwise gains +0.02, so prefer [LightOn-rerank-LW-4B](https://huggingface.co/lightonai/LightOn-rerank-LW-4B) at this size.
 - Training data is English-only. French works zero-shot (the backbone is multilingual) but is slightly behind English on average.
-- Vision hard negatives were mined with ColPali-family retrievers; the model pairs best with a ColPali-family first stage (e.g. ColQwen2.5) for visual reranking.
-- BEIR contamination flag: NQ and MSMARCO are part of the text training data; headline text figures use clean means that exclude them.
+- BEIR contamination flag: NQ and MSMARCO are part of the text training data; headline text figures use decontaminated means that exclude them.
 
 ## The LightOn-rerank family
 
-| Model | Backbone | Scoring | ViDoRe V3 @10 |
+| Model | Backbone | Scoring | ViDoRe V3 overall nDCG@10 |
 |---|---|---|---|
 | [LightOn-rerank-PW-0.8B](https://huggingface.co/lightonai/LightOn-rerank-PW-0.8B) | Qwen3.5-0.8B | pointwise | 0.4820 |
 | [LightOn-rerank-LW-0.8B](https://huggingface.co/lightonai/LightOn-rerank-LW-0.8B) | Qwen3.5-0.8B | listwise | 0.5825 |