INDEX

Explanations

references to events or occurrences that happened in the past

oai_token-act-pair · gpt-3.5-turbo

references to time periods labeled as "past" or similar phrases indicating previous occurrences

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

roots

-0.71

oya

-0.68

wagon

-0.66

 Inher

-0.64

hyde

-0.63

atism

-0.62

cakes

-0.61

obal

-0.61

lees

-0.60

unk

-0.58

POSITIVE LOGITS

decade

1.46

 couple

1.35

few

1.34

ebin

1.33

 fortnight

1.25

 tense

1.20

 month

1.18

 year

1.16

 week

1.15

 century

1.10

Activations Density 0.044%

references to events or occurrences that happened in the past

references to time periods labeled as "past" or similar phrases indicating previous occurrences

No Comments

No Known Activations