INDEX

Explanations

phrases directing the reader to take action or provide input at the end of a text

oai_token-act-pair · gpt-3.5-turbo

references to a "below" section, indicating a request for user engagement or feedback

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 4-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.4.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.4.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

Sense

-0.77

eg

-0.74

ãĤ£

-0.74

oka

-0.72

MM

-0.72

gans

-0.70

imm

-0.69

olid

-0.69

ãĥı

-0.68

olly

-0.67

POSITIVE LOGITS

below

0.84

 below

0.83

ground

0.71

tics

0.70

 tradem

0.68

 veter

0.68

neath

0.67

 summar

0.67

ebin

0.66

 crest

0.64

Activations Density 0.017%

phrases directing the reader to take action or provide input at the end of a text

references to a "below" section, indicating a request for user engagement or feedback

No Comments

No Known Activations