LPM93/teaching-transformers-classical-italian-sonnets

Teaching Transformers to Write Classical Italian Sonnets

This release package contains four selected artifacts intended for publication from a controlled research study of staged literary adaptation of the existing Minerva 7B Instruct model:

Subfolder	Artifact	Selected endpoint
stage1	Full BF16 model after historical/general Italian adaptation	Update 2,065
stage2	Full BF16 model after non-sonnet poetry adaptation	Update 760
stage3	Full BF16 model after V7 sonnet adaptation	Update 120 of 135 planned
dpo_adapter	Rank-8 PEFT LoRA adapter for the exact selected Stage-3 model	Update 61

The three stages form one sequential lineage. They ultimately begin from sapienzanlp/Minerva-7B-instruct-v1.0 at revision d1fc0f0e589ae879c5ac763e0e4206a4d14a3f6d. The models were not trained from scratch. The DPO adapter must be attached to the stage3 subfolder in this repository; it is not compatible with a generic Minerva checkpoint.

Loading

Load each full model by naming its subfolder:

from transformers import AutoModelForCausalLM, AutoTokenizer

repo_id = "LPM93/teaching-transformers-classical-italian-sonnets"

tokenizer = AutoTokenizer.from_pretrained(repo_id, subfolder="stage3")
stage3 = AutoModelForCausalLM.from_pretrained(
    repo_id,
    subfolder="stage3",
    torch_dtype="auto",
    device_map="auto",
)

Load the adapter only after explicitly loading its exact Stage-3 base:

from peft import PeftModel

dpo_model = PeftModel.from_pretrained(
    stage3,
    repo_id,
    subfolder="dpo_adapter",
)

Use subfolder="stage1" or subfolder="stage2" for the earlier full-model states. PEFT's adapter_config.json records the repository ID but has no standard field for a base-model subfolder, so automatic adapter-only loading is not supported: load stage3 explicitly as shown above.

Result boundary

The final sealed test supports a narrow automatic surface/completion gain from AI-judged DPO, not broad literary quality. Human/AI calibration was only 12/20 and failed its gate. Both Stage 3 and DPO produced 0/100 strict-good outputs in the frozen blind review. Fourteen-line output was decoder-controlled. These artifacts are research evidence, not reliable poets, human-aligned systems, or production language models.

Rights and training-data boundary

Hugging Face metadata uses CC BY-NC 4.0 only for copyright and similar rights Leonardo Pacciani-Mori holds, if any, in his original model modifications. Rights independently received in the pinned Minerva parent retain the parent's Apache-2.0 designation. Leonardo-owned model-card prose and aggregate documentation retain CC BY 4.0 where identified.

No corpus text, token streams, prompts, openings, generations, preferences, votes, annotations, mappings, optimizer state, or training logs are included. Source disclosure does not grant corpus-redistribution rights or create one package-wide data license. Read each subfolder's RIGHTS_SCOPE.md, NOTICE.md, TRAINING_CONTENT_SUMMARY.md, and lineage.json before reuse.

The release uses the unresolved assumption that training-data licenses do not govern the weights. If incompatible ShareAlike terms are determined to attach, distribution is not authorized under this structure. The repository is provided without a warranty of title or non-infringement, subject to the included license texts.

Ownership and AI contribution

Leonardo Pacciani-Mori conceived and directed the project, made executive decisions, approved the research plan, reviewed outputs, and sometimes ran GPU work. Codex 5.5 and later Codex 5.6 Sol substantially assisted research design, implementation, tests, execution, and analysis. The study was not independently designed or independently implemented by Leonardo.

Project source, evidence, and full limitations: https://github.com/LeonardoPaccianiMori/portfolio-transformer-poetry