Update README.md

README.md

@@ -1,7 +1,7 @@
 ---
 language:
 - en
-license: gpl-3.0
+license: apache-2.0
 pretty_name: GCI-Bench (Glint Clarity Index Bench)
 task_categories:
 - text-generation
@@ -13,21 +13,17 @@ tags:
 - attention
 ---
 
-# GCI-Bench: Glint Clarity Index Benchmark
+# ACI: Attention Clarity Index Benchmark
 ## Gradient × Attention Context Importance Benchmark
 
-GCI-Bench is a **5,000-item diagnostic benchmark** designed to measure whether a small language model (1M–100M parameters) uses its **attention × gradient interactions** to prioritize relevant context over distractors.
+ACI-Bench is a **5,000-item diagnostic benchmark** designed to measure whether a small language model (1M–150M parameters) uses its **attention × gradient interactions** to prioritize relevant context over distractors.
 
-Unlike standard NLP benchmarks, GCI-Bench does **not evaluate answer correctness**. Instead, it analyzes the model’s internal dynamics while it processes the prompt. Specifically, it measures how strongly gradients flow through attention connections tied to causally relevant parts of the input. No reference answer is required for scoring.
+Unlike standard NLP benchmarks, ACI-Bench does **not evaluate answer correctness**. Instead, it analyzes the model’s internal dynamics while it processes the prompt. Specifically, it measures how strongly gradients flow through attention connections tied to causally relevant parts of the input. No reference answer is required for scoring.
 
 The goal is to evaluate *how* a model distributes importance across context—not whether it produces the right output.
 
 ---
 
-
-### Note
-> Enderchef, the creator of this dataset, has left the org. Permission to this benchmark is given with this credit line: https://huggingface.co/Enderchef
-
 ## Scoring
 
 Each item produces two independent metrics:
@@ -36,7 +32,7 @@ Each item produces two independent metrics:
 |-------|--------|-----------------|
 | **Priority Score** | 0–100 (50 = neutral) | Whether the model assigns more gradient-weighted attention to relevant segments than to distractors |
 | **Linkage Score** | 0–100 (50 = neutral) | Whether the model strengthens attention between causally linked sentence pairs compared to related→distractor pairs |
-| **GCI Score** | 0–100 | Mean of Priority and Linkage |
+| **ACI Score** | 0–100 | Mean of Priority and Linkage |
 
 Both metrics are **self-normalized**. Each item contains a balanced 50/50 split of related and unrelated segments, and scores are averaged per token. A model with no systematic preference will converge to ~50 by construction.
 
@@ -50,7 +46,7 @@ Scores above 50 indicate structured prioritization. Scores below 50 indicate inv
 
 | Field | Type | Description |
 |--------|------|-------------|
-| `id` | string | Unique ID (`gci-00001`–`gci-05000`) |
+| `id` | string | Unique ID (`aci-00001`–`aci-05000`) |
 | `topic` | string | Topic category ID (e.g., `cooking`, `astronomy`) |
 | `topicLabel` | string | Human-readable topic |
 | `templateId` | string | Causal template (`A`–`F`) |
@@ -100,7 +96,7 @@ The harness:
 3. Backpropagates gradients through attention matrices  
 4. Computes attention × gradient attribution  
 5. Aggregates token-level values into segment-level scores  
-6. Produces Priority, Linkage, and final GCI scores  
+6. Produces Priority, Linkage, and final ACI scores  
 
 Example usage:
 
@@ -118,4 +114,4 @@ Some architectures may require minor adjustments to expose attention tensors cor
 
 ## License
 
-GPL 3.0 - see LICENSE
+Apache 2.0 - see LICENSE