---
license: apache-2.0
library_name: transformers
pipeline_tag: text-generation
tags:
- gpt2
- boris
---

# boris-1-125M-first-run

A 12-layer, 768-dim GPT-2-tier (~125M parameter) language model
trained from scratch with the Boris-1 pipeline, exported to `GPT2LMHeadModel`.
Exported from checkpoint `best.pt`.

## Usage

```python
from transformers import AutoModelForCausalLM, AutoTokenizer

tok = AutoTokenizer.from_pretrained("<user>/boris-1-125M")
model = AutoModelForCausalLM.from_pretrained("<user>/boris-1-125M")

ids = tok("The capital of France is", return_tensors="pt").input_ids
print(tok.decode(model.generate(ids, max_new_tokens=40)[0]))
```

## Details

| | |
|---|---|
| Architecture | GPT-2 (pre-LN, learned positional embeddings, tied embeddings) |
| Layers / heads / d_model | 12 / 12 / 768 |
| Context length | 1024 |
| Vocab | 50304 (GPT-NeoX-20B BPE, padded) |
| Tokenizer | `EleutherAI/gpt-neox-20b` |
| Precision | trained in bf16 autocast with fp32 master weights |

Base model; not instruction-tuned or aligned. Outputs may be inaccurate or
offensive.