File size: 2,957 Bytes
638e023 c757d9e 638e023 c757d9e 638e023 c757d9e 638e023 c757d9e 638e023 c757d9e 638e023 c757d9e 638e023 c757d9e 638e023 c757d9e 638e023 c757d9e 638e023 c757d9e 638e023 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 |
---
model_name: Phi-4-Argunaut-1-SFT
license: mit
datasets:
- DebateLabKIT/deepa2-conversations
- DebateLabKIT/deep-argmap-conversations
- allenai/tulu-3-sft-mixture
base_model:
- unsloth/phi-4
pipeline_tag: text-generation
library_name: transformers
tags:
- logic
- argumentation
- critical-thinking
- argument-mapping
- trl
- sft
---
# Model Card for Phi-4-Argunaut-1-SFT
This model is a fine-tuned version of [unsloth/phi-4](https://huggingface.co/unsloth/phi-4).
It has been trained using [TRL](https://github.com/huggingface/trl).
📘 [HF Blog Article](https://huggingface.co/blog/ggbetz/argunauts-phase-1)
[<img src="https://raw.githubusercontent.com/wandb/assets/main/wandb-github-badge-28.svg" alt="Visualize in Weights & Biases" width="150" height="24"/>](https://wandb.ai/ggbetz/Argunauts-1/runs/4b99kqwz/overview)
## Quick start
```python
from transformers import pipeline
question = "Are you familiar with Argdown syntax? What's its purpose?"
generator = pipeline("text-generation", model="DebateLabKIT/Llama-3.1-Argunaut-1-8B-SFT", device="cuda")
output = generator([{"role": "user", "content": question}], max_new_tokens=128, return_full_text=False)[0]
print(output["generated_text"])
```
## Evaluation
### Chat Experience
_coming soon_
### Metrics
_coming soon_
## SFT dataset mixture
|Dataset|Weight (examples)|Weight (tokens)|
|:------|:----:|:----:|
|DebateLabKIT/deepa2-conversations|25%|49%|
|DebateLabKIT/deep-argmap-conversations|25%|18%|
|allenai/tulu-3-sft-mixture|50%|33%|
## Training procedure
Trained with SFT on **1M examples** and for 1 epoch with
* context length 8196
* packing (trl implementation)
* *spectrum* (top 50 percent)
```yaml
# Training parameters
num_train_epochs: 1
per_device_train_batch_size: 2
gradient_accumulation_steps: 8
gradient_checkpointing: true
gradient_checkpointing_kwargs:
use_reentrant: false
learning_rate: 2.0e-6
lr_scheduler_type: cosine
warmup_ratio: 0.1
```
Hardware: 4 x H100 GPUs.
_This work was performed on the HoreKa supercomputer funded by the
Ministry of Science, Research and the Arts Baden-Württemberg and by
the Federal Ministry of Education and Research._
### Framework versions
- TRL: 0.14.0
- Transformers: 4.46.3
- Pytorch: 2.4.1
- Datasets: 3.1.0
- Tokenizers: 0.20.3
## Credits
This work wouldn't be possible without all the **great contributions from the open LLM community**. Thank you! Special kudos go to
- @philschmid for his latest [fine-tuning boilerplate](https://www.philschmid.de/fine-tune-llms-in-2025)
- @lvwerra, @lewtun et al for building and maintaining [trl](https://github.com/huggingface/trl)
- @cognitivecomputations for sharing [spectrum](https://github.com/cognitivecomputations/spectrum/tree/main)
- @allenai for releasing [tulu-3-sft-mixture](https://huggingface.co/datasets/allenai/tulu-3-sft-mixture)
- @microsoft-research for building and @unsloth for recasting [phi-4](https://huggingface.co/microsoft/phi-4)
|