Nehanda
A 7B parameter language model fine-tuned for intelligence assessment, signal detection, and global systems analysis.
Model: asoba/nehanda-v1-7b on Hugging Face
Overview
Nehanda v1 is a specialized language model that departs from standard chat models to focus on forensic analysis and evidence-based assessment. Built on Mistral-7B-v0.3, it prioritizes provenance and structure over fluency, explicitly stating when information is unknown rather than fabricating.
Nehanda serves as the default synthesis engine for Zorora, powering deep research workflows that require rigorous citation tracing and credibility assessment.
Core Capabilities
Signal Detection
Distinguishes between routine noise and pre-cursor indicators of structural shifts in regulatory, financial, and geopolitical systems.
Systems Analysis
Trained on a 10GB corpus spanning regulatory frameworks, financial investigations, and policy doctrines across multiple jurisdictions.
Citation Tracing
Follows logic chains across multiple sources with provenance tracking, enabling verification of claims back to original documents.
Anti-Fabrication
Enforces strict adherence to provided context. Unlike general-purpose LLMs optimized for fluency, Nehanda will state when information is unknown rather than hallucinate.
Training Data
17,852 curated documents (~10GB) across specialized domains:
| Domain | Documents | Purpose |
|---|---|---|
| The Hegemony Layer (USA) | 15,920 | Federal legislation, state policy, administrative ideology |
| The Infrastructure Layer (SA) | 1,559 | Utility whitepapers, grid codes |
| The Systems Layer (Finance) | 213 | Corruption investigations, Panama Papers context, commodities data |
| The Risk Layer (Insurance) | 160 | Risk management textbooks, actuarial logic |
| The Poly Hegemony Layer (Global) | 109 | Non-Western policy doctrines from Russia, China, Brazil, India |
Architecture
| Specification | Value |
|---|---|
| Base Model | Mistral-7B-v0.3 |
| Fine-tuning | LoRA |
| Parameters | 7B |
| Context Window | 4096 tokens |
| Tensor Type | BF16 |
Training Pipeline
- Stage 1: Generic instruction following + strict logic/reasoning (math-hardened)
- Stage 2: Systems knowledge via 10GB contextual ingestion
- Stage 3: Signal persona training for intelligence assessment Q&A
Integration with Zorora
Nehanda powers the /search and /research commands in Zorora:
- Ingest - Raw context from search tools
- Triage - Information scored by credibility and relevance
- Synthesize - Answers highlighting information gaps, conflicting accounts, and consensus points
Usage
from transformers import AutoModelForCausalLM, AutoTokenizer
model_id = "asoba/nehanda-v1-7b"
tokenizer = AutoTokenizer.from_pretrained(model_id)
model = AutoModelForCausalLM.from_pretrained(model_id, load_in_4bit=True)
prompt = """You are an intelligence assessment specialist.
### Instruction:
Analyze the provided cable for indicators of regulatory capture.
### Context:
[Your context here]
### Response:"""
inputs = tokenizer(prompt, return_tensors="pt")
outputs = model.generate(**inputs, max_new_tokens=512)
print(tokenizer.decode(outputs[0]))
GGUF Version
A quantized version is available for efficient local inference: asoba/nehanda-v1-7b-GGUF
License
CC-BY-NC-ND-4.0
Access requires contact information sharing via Hugging Face.
Support & Resources
Documentation
Support
- Email: support@asoba.co
- Discord: Join our community
Get Help & Stay Updated
Contact Support
For technical assistance, feature requests, or any other questions, please reach out to our dedicated support team.
Email Support Join Discord