Update README.md

README.md

@@ -35,6 +35,8 @@ configs:
 ---
 
 # Multimodal Pretraining
+This section covers our large-scale collections at the source and is distributed in its original form — PDF with layout intact, audio attached to its transcript — rather than as text extracted after the fact. The emphasis is on what large-scale web collection  misses: academic global production (badly indexed in scientific repositories); patents outside the US; the technical and regulatory archives of telecom and finance. 
+These are long, structured, image-dense documents, which makes them a good fit for long-context and long-horizon multimodal training. Samples mirror the structure of the full assets.
 
 ## Global academic thesis
 
@@ -92,6 +94,9 @@ The current sample includes 5,000 documents in PDF, spanning every region, with
 
 # Synthetic data seeds/environment
 
+We provide synthetic datasets and environments grounded in open or licenced proprietary seed material, the specifications and the pipelines needed to generate at scale. Each environment is anchored in a real structured asset such demographic distributions, knowledge graphs, etc. 
+Our NeurIPS 2026 paper measures what this is worth. SYNTH — ~80B tokens generated from 58,000 seed articles — trains models that sit on the token-efficiency frontier with no SFT or RL stage at all: within 0.7 points of Qwen3-0.6B on open-ended tasks at 80–700× fewer training tokens, and best-in-tier factual precision at 10–140× fewer tokens.
+
 ## Global personas.
 
 Our internal assets scaling the synthetic personas at a global scale. We collected a unique corpus of representative first name and last names and aggregated many demographics distribution from international organizations and academic research.