INDEX

Explanations

terms related to video games, such as levels, boss fights, rewards, achievements, missions, enemies, and unlocking content

oai_token-act-pair · gpt-3.5-turbo

references to boss fights in gaming contexts

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 4-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.4.hook_resid_pre

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

Data Type

torch.float32

Hook Point

blocks.4.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

internet

-0.83

forth

-0.79

hem

-0.73

 hemorrh

-0.71

rup

-0.70

noon

-0.70

 lashed

-0.69

 politic

-0.69

olia

-0.68

maid

-0.68

POSITIVE LOGITS

 quests

1.05

 playthrough

1.02

 Chocobo

1.01

 dungeons

1.00

 Reward

0.99

 unlocks

0.96

 Crafting

0.94

 Unlock

0.94

Reward

0.94

ocobo

0.93

Activations Density 0.200%

terms related to video games, such as levels, boss fights, rewards, achievements, missions, enemies, and unlocking content

references to boss fights in gaming contexts

No Comments

No Known Activations