INDEX

Explanations

phrases related to events or actions coming to a halt or being interrupted

oai_token-act-pair · gpt-3.5-turbo

instances of the word "stopped" in relation to various subjects

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

rocket

-0.80

ighth

-0.76

aths

-0.74

Space

-0.74

é¾įåĸļå£«

-0.72

elf

-0.71

Sov

-0.70

edded

-0.69

olesc

-0.69

ospace

-0.68

POSITIVE LOGITS

 bothering

0.83

gap

0.81

watch

0.78

watching

0.75

 abruptly

0.75

 smoking

0.74

ŃĶ

0.71

 blinking

0.68

 seiz

0.66

 ringing

0.65

Activations Density 0.031%

phrases related to events or actions coming to a halt or being interrupted

instances of the word "stopped" in relation to various subjects

No Comments

No Known Activations