---
library_name: transformers
pipeline_tag: text-generation
---

# TorchWright Doom — E1M1

This is a stock Hugging Face `Phi3ForCausalLM` that renders DOOM through
ordinary autoregressive inference. The model and the data-only fast tokenizer
load through the ordinary Transformers text-generation pipeline without
remote code.

The bundled `examples/e1m1_prompt.txt` is the executable prompt. Run
`infer.py` (at the bundle root) to produce canonical emitted row ids and raw
tokenizer text. `tools/pretty_text.py` formats that text for reading, while
`tools/txt_to_png.py` independently decodes its cursor/pixel protocol into a
PNG — every cursor move and pixel in that protocol is a model-emitted token.
The protocol is specified in `PROTOCOL.md` in the source repo. Neither
post-processing tool participates in inference or performs geometry,
visibility, lighting, texture selection, or sorting.

Ordinary Transformers pipeline inference works directly, with no custom or
remote model code:

```python
from pathlib import Path
from huggingface_hub import hf_hub_download
from transformers import pipeline

repo = "physicsrob/torchwright-doom-e1m1"
prompt = Path(hf_hub_download(repo, "examples/e1m1_prompt.txt")).read_text()
generate = pipeline("text-generation", model=repo, device_map="auto")
generated_text = generate(prompt, return_full_text=False)[0]["generated_text"]
```

The saved generation defaults are greedy and cover the complete frame. Use
the shipped `infer.py` when canonical integer row IDs, progress reporting, and
the exact terminal-token-preserving raw text are required.

Published checkpoints: [320×200](https://huggingface.co/physicsrob/torchwright-doom-e1m1)
and [80×50](https://huggingface.co/physicsrob/torchwright-doom-e1m1-80x50).
The compiler-facing source is
[torchwright_doom](https://github.com/physicsrob/torchwright_doom).

**This bundle:** screen 320×200, map
E1M1, dense fp32 sharded safetensors, eager attention (the validated
implementation), greedy decode, generation bound
61440 new tokens.

**What running it takes:** the fp32 weight shards total
79.97 GiB (85.87 GB), needing a B200-class GPU
or multi-GPU `device_map`. The flagship pipeline render peaked at 151.00 GiB
reserved; greedy decode took 39.7 minutes on one B200 for its 53,747-token rollout
from a 3,614-token prompt, scoring 99.9% within-option color against the
reference renderer.

Canonical numbers and their provenance: `FACTS.md` in the source repo
(github.com/physicsrob/torchwright_doom).