INDEX

Explanations

phrases that indicate enjoyment or fun

oai_token-act-pair · gpt-3.5-turbo

instances of the word "fun" in various contexts

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

 bottleneck

-0.72

 Centauri

-0.69

BOOK

-0.63

attle

-0.62

 underest

-0.61

 proport

-0.61

obook

-0.59

 Transparency

-0.59

 helicop

-0.59

 defective

-0.59

POSITIVE LOGITS

nels

1.57

gal

1.14

eral

1.09

nell

1.09

nel

1.02

ctor

0.93

ctory

0.86

ilee

0.86

ctions

0.83

imation

0.83

Activations Density 0.025%

phrases that indicate enjoyment or fun

instances of the word "fun" in various contexts

No Comments

No Known Activations