Update paper title for clarity and specificity

dariocazzani · dariocazzani · commit 8dbb9840a48e · 2026-03-14T17:07:15.000-04:00
- Add "1.58-bit" for concrete precision (vs vague "ternary")
- Add "in CNNs" to clarify scope (not transformers/LLMs)
- Update paper/main.tex title and LaTeX comment
- Update README.md heading and BibTeX citation
- Update pyproject.toml description for consistency
- Recompile paper with new title (28 pages, 553 KB)

New title: "Understanding and Closing the 1.58-bit
Quantization Gap in CNNs: An Empirical Study"

Better for TMLR submission:
- More searchable ("1.58-bit" + "CNNs" are concrete keywords)
- Prevents confusion with BitNet LLM work
- Combines problem focus with technical specificity
diff --git a/README.md b/README.md
@@ -1,4 +1,4 @@
-# 1.58-bit Convolutional Neural Networks: A Systematic Comparison Study
+# Understanding and Closing the 1.58-bit Quantization Gap in CNNs: An Empirical Study
 
 > Research reproducibility repository for TMLR submission
 
@@ -295,7 +295,7 @@ bitnet/
 
 ```bibtex
 @article{cazzani2026bitconv,
-  title={1.58-bit Convolutional Neural Networks: A Systematic Comparison Study},
+  title={Understanding and Closing the 1.58-bit Quantization Gap in CNNs: An Empirical Study},
   author={Cazzani, Dario},
   year={2026}
 }
diff --git a/paper/main.pdf b/paper/main.pdf
diff --git a/paper/main.tex b/paper/main.tex
@@ -1,4 +1,4 @@
-% TMLR Submission: Understanding and Closing the Ternary Quantization Gap
+% TMLR Submission: Understanding and Closing the 1.58-bit Quantization Gap in CNNs
 % New version with modular sections
 
 \documentclass[10pt]{article}
@@ -17,7 +17,7 @@
 % METADATA
 % =============================================================================
 
-\title{Understanding and Closing the Ternary Quantization Gap:\\An Empirical Study}
+\title{Understanding and Closing the 1.58-bit Quantization Gap in CNNs:\\An Empirical Study}
 
 \author{
   Dario Cazzani\\
diff --git a/pyproject.toml b/pyproject.toml
@@ -1,7 +1,7 @@
 [project]
 name = "bitnet-1.58"
 version = "0.1.0"
-description = "1.58-bit Neural Networks: Systematic Comparison Study"
+description = "Understanding and Closing the 1.58-bit Quantization Gap in CNNs: An Empirical Study"
 requires-python = ">=3.11"
 dependencies = [
     "torch>=2.0.0",
