INDEX

Explanations

text related to Twilight series

oai_token-act-pair · gpt-3.5-turbo

occurrences of the token "tw" 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 @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

 PRESS

-0.79

HAEL

-0.73

ulative

-0.72

senal

-0.67

mble

-0.66

ãĤ¹ãĥĪ

-0.65

MIC

-0.65

ãĤŃ

-0.64

Journal

-0.63

-0.62

POSITIVE LOGITS

enty

1.08

tw

1.01

elve

0.96

orks

0.92

ymes

0.82

eteen

0.81

itness

0.78

ixt

0.78

actory

0.77

Tw

0.75

Activations Density 0.006%

text related to Twilight series

occurrences of the token "tw" in various contexts

No Comments

No Known Activations