94 lines
2.4 KiB
Markdown
94 lines
2.4 KiB
Markdown
|
---
|
||
|
|
language:
|
||
|
|
- en
|
||
|
|
- it
|
||
|
|
license: apache-2.0
|
||
|
|
tags:
|
||
|
|
- text-generation
|
||
|
|
- gpt2
|
||
|
|
- decoder-only
|
||
|
|
- causal-lm
|
||
|
|
- distillation
|
||
|
|
- dpo
|
||
|
|
- sft
|
||
|
|
- lora
|
||
|
|
library_name: transformers
|
||
|
|
pipeline_tag: text-generation
|
||
|
|
model-index:
|
||
|
|
- name: FF_3.1
|
||
|
|
results:
|
||
|
|
- task:
|
||
|
|
type: text-generation
|
||
|
|
name: Text Generation
|
||
|
|
dataset:
|
||
|
|
name: MMLU
|
||
|
|
type: cais/mmlu
|
||
|
|
metrics:
|
||
|
|
- type: accuracy
|
||
|
|
value: 27.94
|
||
|
|
name: 5-shot accuracy
|
||
|
|
---
|
||
|
|
|
||
|
|
# FF_3.1
|
||
|
|
|
||
|
|
**FF_3.1** is a 2.02B parameter GPT-2 decoder-only language model trained from scratch with a multi-stage pipeline combining supervised fine-tuning, preference optimization, knowledge distillation, and instruction tuning.
|
||
|
|
|
||
|
|
## Model Details
|
||
|
|
|
||
|
|
| | |
|
||
|
|
|---|---|
|
||
|
|
| **Architecture** | GPT-2 decoder-only |
|
||
|
|
| **Parameters** | 2.02B |
|
||
|
|
| **Hidden size (d)** | 2048 |
|
||
|
|
| **Attention heads (h)** | 16 |
|
||
|
|
| **FFN size (ff)** | 8192 |
|
||
|
|
| **Layers (L)** | 38 |
|
||
|
|
| **Context length** | 2048 |
|
||
|
|
| **Tokenizer** | GPT-2 BPE (vocab size: 50,257) |
|
||
|
|
| **Precision** | bfloat16 |
|
||
|
|
|
||
|
|
## Training Pipeline
|
||
|
|
|
||
|
|
FF_3.1 was trained through a 5-stage pipeline:
|
||
|
|
|
||
|
|
1. **Pretraining** — 90B tokens on a large English corpus
|
||
|
|
2. **SFT** — 760K + 100K examples (OpenHermes-2.5 / NuminaMath / Eurus)
|
||
|
|
3. **DPO** — 38,863 preference pairs
|
||
|
|
4. **Distillation v3** — 47K examples targeting MMLU + GSM8K + ARC benchmarks
|
||
|
|
5. **LoRA v4b** — 10K examples for instruction following refinement
|
||
|
|
|
||
|
|
## Evaluation
|
||
|
|
|
||
|
|
| Benchmark | Score |
|
||
|
|
|---|---|
|
||
|
|
| **MMLU (5-shot)** | **27.94%** (+3.94 pp vs FF_3 baseline of 24%) |
|
||
|
|
|
||
|
|
## Usage
|
||
|
|
|
||
|
|
```python
|
||
|
|
from transformers import AutoModelForCausalLM, AutoTokenizer
|
||
|
|
|
||
|
|
model = AutoModelForCausalLM.from_pretrained("francescofiamingo1/FF_3.1", torch_dtype="bfloat16")
|
||
|
|
tokenizer = AutoTokenizer.from_pretrained("francescofiamingo1/FF_3.1")
|
||
|
|
|
||
|
|
input_text = "Explain photosynthesis in simple terms."
|
||
|
|
inputs = tokenizer(input_text, return_tensors="pt").to(model.device)
|
||
|
|
outputs = model.generate(**inputs, max_new_tokens=256, temperature=0.7, do_sample=True)
|
||
|
|
print(tokenizer.decode(outputs[0], skip_special_tokens=True))
|
||
|
|
```
|
||
|
|
|
||
|
|
## Known Limitations
|
||
|
|
|
||
|
|
- **Math reasoning** is still weak — the model struggles with multi-step arithmetic and word problems
|
||
|
|
- **Instruction count following** is imprecise — the model may not reliably follow constraints like "list exactly 5 items"
|
||
|
|
|
||
|
|
## What's Next
|
||
|
|
|
||
|
|
**FF_3.2** will focus on:
|
||
|
|
- DPO with UltraFeedback dataset for improved preference alignment
|
||
|
|
- Improved math dataset for stronger quantitative reasoning
|
||
|
|
|
||
|
|
## License
|
||
|
|
|
||
|
|
Apache 2.0
|