Update readme

README.md

@@ -20,14 +20,15 @@ A normalized semi-synthetic Python dataset for training small language models on
 
 ![Tokenizer chart](tokenizer_chart.png)
 
-## Why NPset-2?
+## Why
+
+Small language models trained on natural language corpora develop latent representations of logical constructs -- iteration, conditionals, data flow, function composition -- yet struggle to apply this reasoning to source code, where syntactic overhead (delimiters, indentation conventions, language-specific idioms) occupies a disproportionate share of the token budget, requires a vocabulary of code-specific tokens, and introduces a surface-form distribution shift relative to the model's prior knowledge. NPset-2 addresses this by normalizing Python source through an AST-based converter that strips syntactic noise while preserving the full logical structure of each program, producing a pseudocode representation composed entirely of natural language tokens that aligns more directly with the semantic representations already present in small models, allowing them to reason about what code *does* rather than expending capacity learning what it *looks like*.
 
-Small language models develop latent logical representations but struggle with syntactic overhead. NPset-2 addresses this through the **TinyDSL v2** specification, which strips syntactic noise and provides explicit logical anchors. 
 
 ## The v2 Specification
 
+NPset-2 introduces significant improvements over v1, trading some relative token compression for far lower semantic overhead.
 
-NPset-2 introduces significant improvements over v1, designed to minimize "token tax" and maximize the model's ability to track long-range logical dependencies:
 
 1.  **Explicit Block Scoping**: All indented blocks (if, for, while, try, with) now use numbered/named anchors: `begin if 1` ... `end if 1`. This provides unambiguous attention anchors for small models.
 2.  **Natural Language Phrasing**:
@@ -43,14 +44,14 @@ NPset-2 introduces significant improvements over v1, designed to minimize "token
 
 ## Performance (Context Capacity)
 
-When tested against standard tokenizers, TinyDSL v2 significantly expands the effective context window for logic-heavy training:
+When tested against standard tokenizers, TinyDSL v2 significantly expands the effective context window for logic-heavy training with natural language tokenizers:
 
 | Tokenizer | Reduction (Tokens) | Context Capacity (2048 window) |
 | :--- | :--- | :--- |
-| **GPTX (Custom 32k)** | **+13.7%** | **7.1 -> 8.3 examples (+15.9%)** |
-| **GPT-2** | **+16.6%** | **7.4 -> 8.9 examples (+19.9%)** |
-| Qwen 2.5 | -8.1% | 10.1 -> 9.3 examples (-7.5%) |
-| Llama 3 | -2.2% | 8.3 -> 8.1 examples (-2.2%) |
+| **GPTX (Custom 32k)** | **13.7%** | **7.1 -> 8.3 examples (+15.9%)** |
+| **GPT-2** | **16.6%** | **7.4 -> 8.9 examples (+19.9%)** |
+| Qwen 2.5 | 8.1% | 10.1 -> 9.3 examples (-7.5%) |
+| Llama 3 | 2.2% | 8.3 -> 8.1 examples (-2.2%) |
 
 *Note: While raw character counts increase by ~17%, the "Token Tax" for logical constructs is drastically reduced for models not pre-specialized for code syntax.*
 
@@ -67,9 +68,5 @@ Parquet format with the following schema:
 
 ## Sources
 
-The dataset is compiled from high-quality educational and stack-overflow style Python sources, including:
-- `dbands/pythonMath`
-- `nomic-ai/cornstack-python-v1`
-- `zaydzuhri/stack-edu-python`
-- `jtatman/python-code-dataset-500k`
-- And other curated sources.
+- `HuggingFaceTB/stack-edu (python)`
+