INDEX

Explanations

negation cues

oai_token-act-pair · gpt-3.5-turbo

negations or words indicating something not occurring

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.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

icka

-0.64

anza

-0.64

arthed

-0.63

pmwiki

-0.62

cies

-0.61

DAQ

-0.61

UE

-0.61

shore

-0.61

song

-0.59

bys

-0.55

POSITIVE LOGITS

not

2.35

NOT

1.88

not

1.88

Not

1.60

Not

1.56

NOT

1.38

 never

1.25

 neither

1.17

nt

1.14

 indeed

1.07

Activations Density 0.217%

negation cues

negations or words indicating something not occurring

No Comments

No Known Activations