INDEX

Explanations

phrases mentioning certain acts or intentions related to legal or criminal activities

oai_token-act-pair · gpt-3.5-turbo

references to possession or involvement in actions

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 11-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.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.11.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

hooting

-0.79

etter

-0.72

rift

-0.64

osi

-0.64

anding

-0.64

undown

-0.63

umping

-0.63

iscovery

-0.62

raft

-0.62

ourgeois

-0.61

POSITIVE LOGITS

 been

1.20

 participated

1.17

 acted

1.14

 undergone

1.09

 complied

1.08

 permission

1.05

been

1.04

 behaved

1.04

 violated

1.04

 resided

1.04

Activations Density 0.234%

phrases mentioning certain acts or intentions related to legal or criminal activities

references to possession or involvement in actions

No Comments

No Known Activations