INDEX

Explanations

log-in related actions and prompts

oai_token-act-pair · gpt-3.5-turbo

phrases related to logging into accounts or services

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

catentry

-0.83

 gelatin

-0.74

 focal

-0.68

OIL

-0.65

ä½¿

-0.65

tal

-0.65

ĸļ

-0.64

Grade

-0.63

 gestation

-0.62

 coating

-0.61

POSITIVE LOGITS

 Username

0.76

reens

0.73

ername

0.72

ministic

0.70

 Register

0.70

ulhu

0.70

oshenko

0.69

anian

0.69

redit

0.69

 subscrib

0.68

Activations Density 0.063%

log-in related actions and prompts

phrases related to logging into accounts or services

No Comments

No Known Activations