INDEX

Explanations

expressions indicating that something is incorrect or problematic

oai_token-act-pair · gpt-3.5-turbo

phrases indicating the absence or presence of wrongdoing

oai_token-act-pair · gpt-4o-mini

Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 11-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.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.11.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Negative Logits

quickShipAvailable

-0.70

 Hemp

-0.65

ADRA

-0.63

NetMessage

-0.62

hyde

-0.61

prise

-0.60

 Rudd

-0.58

akeru

-0.57

ultz

-0.57

ilt

-0.56

POSITIVE LOGITS

fully

0.92

eous

0.79

headed

0.79

ful

0.77

doing

0.74

with

0.72

 with

0.71

nered

0.71

 WITH

0.70

lihood

0.69

Activations Density 0.044%

expressions indicating that something is incorrect or problematic

phrases indicating the absence or presence of wrongdoing

No Known Activations