LitGram 1.5B LoRA v1.3 - Accuracy Patch

A targeted LoRA adapter for Navyaforaa/LitGram-1.5B.

What it targets

This adapter was trained on a small, verified corrective dataset to improve:

  • restraint about unverified creator, institution, training, and citation claims;
  • attribution of Better to reign in Hell, than serve in Heav'n. to Satan in John Milton's Paradise Lost, Book I, line 263;
  • the 1667 ten-book and revised 1674 twelve-book editions of Paradise Lost;
  • a regular iambic-pentameter scansion of Shall I compare thee to a summer's day?;
  • refusal to fabricate literary citations.

Load with PEFT

Use the base model Navyaforaa/LitGram-1.5B with this adapter through PeftModel.from_pretrained.

Validation

Five held-out paraphrased checks passed after training: provenance restraint, Milton attribution, Paradise Lost edition history, scansion, and refusal to fabricate a citation.

Limitations

This is a narrow corrective adapter, not a general literature benchmark. Verify quotations and scholarly claims against reliable editions.

Downloads last month
-
Inference Providers NEW
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support

Model tree for Navyaforaa/LitGram-1.5B-LoRA-v1.3

Unable to build the model tree, the base model loops to the model itself. Learn more.