INDEX

Explanations

mentions of the word "truth."

oai_token-act-pair · gpt-3.5-turbo

occurrences of the word "truth."

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

 Wolves

-0.68

zzi

-0.68

ATIONS

-0.66

 Apocalypse

-0.65

hiba

-0.64

backer

-0.63

 Zombie

-0.62

 Goods

-0.61

 Attention

-0.60

eland

-0.59

POSITIVE LOGITS

anasia

1.10

osate

0.94

lessness

0.94

lessly

0.94

urst

0.93

ilater

0.90

ouse

0.89

umbing

0.89

sburg

0.88

reys

0.88

Activations Density 0.042%

mentions of the word "truth."

occurrences of the word "truth."

No Comments

No Known Activations