INDEX

Explanations

terms related to terrorism and related activities

oai_token-act-pair · gpt-3.5-turbo

references to terrorism and related legal consequences

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.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

 Saturn

-0.66

 Centauri

-0.65

 Hubble

-0.65

 spheres

-0.65

 needles

-0.63

 saturation

-0.63

 jewelry

-0.62

 electronics

-0.61

 currency

-0.61

 Addiction

-0.61

POSITIVE LOGITS

ists

0.93

emouth

0.91

raq

0.90

ivation

0.90

iffs

0.89

ãĥ³ãĤ¸

0.88

selves

0.87

rats

0.86

icides

0.86

ileaks

0.85

Activations Density 0.040%

terms related to terrorism and related activities

references to terrorism and related legal consequences

No Comments

No Known Activations