upiter commited on
Commit
aee7d41
·
verified ·
1 Parent(s): d6f5e39

Update README.md

Browse files

LangHack is a dataset of [diff history](https://diffhistory.github.io/) demonstration data for the rogue-like video game [NetHack](https://github.com/facebookresearch/nle) generated using the symbolic [AutoAscend bot](https://github.com/maciej-sypetkowski/autoascend), which boasts state-of-the-art performance in the game (as of 07/22/2024).

This dataset was created by sub-sampling 10,000 full NetHack games played by AutoAscend into contiguous "chunks" of 64 timesteps, and converting the agent's game state observations in natural language text using the [NetHack Language Wrapper](https://github.com/ngoodger/nle-language-wrapper). Sub-sampling was performed uniformly at random over all recorded game data.

LangHack prompts correspond to a full game state observation at one timestep of AutoAscend gameplay, while completions correspond to a interleaved set of the subsequent bot actions and their resultant text deltas in the world state.

A detailed report of NetHack agent performance achieved by finetuning a tiny LLM ([GPT2-127M](https://huggingface.co/openai-community/gpt2)) on LangHack is provided [here](https://arxiv.org/abs/2312.07540).

Files changed (1) hide show
  1. README.md +28 -20
README.md CHANGED
@@ -1,20 +1,28 @@
1
- ---
2
- license: mit
3
- dataset_info:
4
- features:
5
- - name: prompt
6
- dtype: string
7
- - name: completion
8
- dtype: string
9
- splits:
10
- - name: train
11
- num_bytes: 13449668588
12
- num_examples: 500000
13
- download_size: 3251708048
14
- dataset_size: 13449668588
15
- configs:
16
- - config_name: default
17
- data_files:
18
- - split: train
19
- path: data/train-*
20
- ---
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ license: mit
3
+ dataset_info:
4
+ features:
5
+ - name: prompt
6
+ dtype: string
7
+ - name: completion
8
+ dtype: string
9
+ splits:
10
+ - name: train
11
+ num_bytes: 13449668588
12
+ num_examples: 500000
13
+ download_size: 3251708048
14
+ dataset_size: 13449668588
15
+ configs:
16
+ - config_name: default
17
+ data_files:
18
+ - split: train
19
+ path: data/train-*
20
+ task_categories:
21
+ - text-generation
22
+ tags:
23
+ - nethack
24
+ - interactive decision-making
25
+ - llm agents
26
+ - imitation learning
27
+ - behavioral cloning
28
+ ---