INDEX

Explanations

mention of a specific news outlet

oai_token-act-pair · gpt-3.5-turbo

occurrences of the word "Post."

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

IBLE

-0.83

oult

-0.66

OTS

-0.65

istered

-0.65

anyon

-0.64

ked

-0.62

paces

-0.62

èª

-0.61

ãĤ¤ãĥĪ

-0.61

FTWARE

-0.61

POSITIVE LOGITS

gres

1.29

greSQL

1.25

erity

1.15

doctoral

1.03

mortem

1.02

modern

0.88

graduate

0.87

natal

0.86

hum

0.83

office

0.81

Activations Density 0.026%

mention of a specific news outlet

occurrences of the word "Post."

No Comments

No Known Activations