INDEX

Explanations

mentions or discussions related to plots or schemes, especially those involving deceit or malicious intent

oai_token-act-pair · gpt-3.5-turbo

references to secretive planning or conspiratorial actions

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

uther

-0.80

inction

-0.77

Kle

-0.76

odor

-0.76

anan

-0.75

osate

-0.73

phal

-0.73

ijk

-0.73

 Columb

-0.73

ported

-0.72

POSITIVE LOGITS

 plotting

1.42

 plotted

1.11

 sche

0.96

 plots

0.91

 conspiring

0.90

 intrig

0.82

heet

0.80

 Plot

0.77

ank

0.74

 blackmail

0.73

Activations Density 0.026%

mentions or discussions related to plots or schemes, especially those involving deceit or malicious intent

references to secretive planning or conspiratorial actions

No Comments

No Known Activations