INDEX

Explanations

the word "watch" or its variations

oai_token-act-pair · gpt-3.5-turbo

instances of the word "watch" in various forms, indicating a focus on observing or viewing events

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 7-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.7.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.7.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

interstitial

-0.84

esan

-0.84

enture

-0.75

ãĥ´

-0.71

wealth

-0.70

ctrl

-0.70

 congr

-0.68

 lettuce

-0.66

inia

-0.63

bole

-0.61

POSITIVE LOGITS

dogs

1.11

 Watching

1.10

tower

1.06

dog

1.05

 watching

0.90

tv

0.84

 helpless

0.84

watching

0.84

 passively

0.79

 replay

0.78

Activations Density 0.042%

the word "watch" or its variations

instances of the word "watch" in various forms, indicating a focus on observing or viewing events

No Comments

No Known Activations