INDEX

Explanations

words related to official approval or authorization

oai_token-act-pair · gpt-3.5-turbo

terms related to punitive measures or restrictions

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

OTE

-0.72

Boh

-0.72

hire

-0.72

irrel

-0.71

lycer

-0.70

_>

-0.66

ophe

-0.65

 Norn

-0.64

Hop

-0.63

thora

-0.63

POSITIVE LOGITS

 sanction

1.01

 sanctions

0.89

 levied

0.84

 sanctioned

0.82

ably

0.79

iless

0.75

 imposed

0.74

SHIP

0.71

AFTA

0.70

iate

0.69

Activations Density 0.012%

words related to official approval or authorization

terms related to punitive measures or restrictions

No Comments

No Known Activations