INDEX

Explanations

references to villains in various contexts

oai_token-act-pair · gpt-3.5-turbo

references to villains and their destructive traits or actions

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

aeda

-0.65

 HELP

-0.63

 authenticated

-0.61

 OFFIC

-0.60

 coerc

-0.59

ERSON

-0.59

 livest

-0.58

 cooperative

-0.55

ITNESS

-0.55

 earners

-0.54

POSITIVE LOGITS

ous

2.84

ously

2.58

OUS

1.77

osity

1.43

uously

1.35

ized

1.32

izing

1.31

ising

1.29

istic

1.27

iously

1.25

Activations Density 0.096%

references to villains in various contexts

references to villains and their destructive traits or actions

No Comments

No Known Activations