Downloads · 30 days
0
phi3.5-fstar-lora
dassarthak18/phi3.5-fstar-lora
phi3.5-fstar-lora is a text generation model from dassarthak18. Use it when you need the model to write or continue text. It is set up for sfttrainer. The card lists the license as mit.
This model is a LoRA fine-tuned version of Microsoft's Phi-3.5-mini-instruct. It specializes in generating F code, including function definitions, lemmas, and formal proofs. In their paper Towards Neural Synthesis for…
Access
Public
Updated Nov 24, 2025
Repo size
83.4 MB
Likes
0
Public
Repo makeup
Click a slice to open those files.
.safetensors71.3 MB · 82%
At a glance
- Task
- Text Generation
- Library
- sfttrainer
- License
- mit
- Access
- Public
- Created
- Nov 24, 2025
- Updated
- Nov 24, 2025
- SHA
- 9cc52b2c
Try a prompt
Base models
Model card
From the Hugging Face model README
Phi-3.5-mini-instruct fine-tuned on F*
This model is a LoRA fine-tuned version of Microsoft's Phi-3.5-mini-instruct. It specializes in generating F* code, including function definitions, lemmas, and formal proofs. In their paper Towards Neural Synthesis for SMT-Assisted Proof-Oriented Programming, Saikat Chakraborty et al. (2024) use Phi-2, among other LLMs, for evaluating their dataset.
Model Details
The model was fine-tuned on the Singularity HPC Cluster at IIT Gandhinagar using a single 48GB NVIDIA L40S GPU. The fine-tuning scripts and console logs are available at scripts directory.
Model Description
- Base Model: microsoft/Phi-3.5-mini-instruct
- Fine-Tuning Dataset: dassarthak18/FStarDataset-V2-Conversation
Evaluation
To examine the tensorboard logs for further training and eval details, use the following commands:
pip install tensorboard
tensorboard --logdir ./logs
Testing Data & Metrics
- Test Loss: 0.9387
- Test Perplexity: 2.56
- Task
- Text Generation
- Library
- sfttrainer
- License
- mit
- Languages
- en
- Created
- Nov 24, 2025
- Updated
- Nov 24, 2025