INDEX

Explanations

statements indicating a final event or additional information

oai_token-act-pair · gpt-3.5-turbo

phrases indicating the concept of finality or repetition

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

bots

-0.76

lees

-0.74

ustomed

-0.71

JS

-0.70

ences

-0.66

idents

-0.65

nis

-0.65

yond

-0.64

orig

-0.64

quire

-0.63

POSITIVE LOGITS

 workaround

0.82

 hurdle

0.76

 slice

0.75

 twist

0.74

 piece

0.74

 flick

0.74

 complication

0.73

 parcel

0.73

 avenue

0.73

gri

0.73

Activations Density 0.070%

statements indicating a final event or additional information

phrases indicating the concept of finality or repetition

No Comments

No Known Activations