---

language:

• en
• de

license: apache-2.0

tags:

• text-generation-inference
• transformers
• unsloth
• llama
• trl
• orpo

base_model: cstr/phi-3-orpo-v8_16

---

Model details

These are the q4 in GGUF of a quick experiment on llamafied phi-3 with only 1000 orpo steps from an azureml translated german orca binarized-dataset (johannhartmann/mistralorpo), with original phi-3 prompt template. The immediate result is not really good, but also not bad enough to disencourage further experiments.

Benchmark results

This was an experiment on a german dataset snippet which, as expected, worsened results on english benchmarks:

| Metric |Value|

|---------------------------------|----:|

|Avg. |64.40|

|AI2 Reasoning Challenge (25-Shot)|60.41|

|HellaSwag (10-Shot) |78.37|

|MMLU (5-Shot) |65.26|

|TruthfulQA (0-shot) |49.76|

|Winogrande (5-shot) |70.24|

|GSM8k (5-shot) |62.32|

On german EQ-Bench (v2_de) 51.82 (insignificant over 51.41 for original llamafied but significantly better than intermediate cstr/phi-3-orpo-v8_16 which after initial 150 test steps achieved 46.38) but with still only 164/171 correctly parsed.

Note: We can improve the correctness of parsing, i.a., by only a few SFT steps, as shown with cas/phi3-mini-4k-llamafied-sft-v3 (170/171 correct but with then only 39.46 score in v2_de, which was also an experiment in changing the prompt template).

All that was quickly done with bnb and q4 quants only, which might, in theory, affect especially such small dense models significantly.

But it served the intention for both proof-of-concept-experiments at least. Probably it would easily be possible to further improve results, but that would take some time and compute.

Training setup

This llama model was trained 2x faster with Unsloth and Huggingface's TRL library.

Provenance and EU AI Act Art. 53 note

• Base model: cstr/phi-3-orpo-v9_16 — a ORPO fine-tune published by the same maintainer as this repository. It is not a third-party upstream: the maintainer authored that model.
• What was done here: format conversion and/or quantisation of that base model only (GGUF). No further training, fine-tuning or merging was applied at this step.
• Licence: apache-2.0, inherited through the base model from the models it was built from.
• Training data: none was used, added or selected at this conversion step. The base model's card lists the models it was built from; their training content is documented — where it is documented at all — by their respective providers.
• Provider status: under Regulation (EU) 2024/1689 this repository makes no provider claim for the conversion step. Any provider obligations attaching to the model itself — including Art. 53(1)(c) copyright policy and Art. 53(1)(d) training-content summary — attach at cstr/phi-3-orpo-v9_16, not here.