INDEX

Explanations

the word "one" followed by a preposition or comma

oai_token-act-pair · gpt-3.5-turbo

phrases indicating sequences or repetitions of actions

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

Ü

-0.88

ESE

-0.74

 Bunker

-0.67

CHR

-0.65

RF

-0.64

 unden

-0.64

 Fighters

-0.62

andise

-0.62

ÃįÃį

-0.62

 GEAR

-0.62

POSITIVE LOGITS

ettings

0.72

acea

0.67

ieth

0.67

teenth

0.65

uden

0.65

shire

0.65

Ã¡s

0.64

othy

0.64

hani

0.64

 neutron

0.64

Activations Density 0.194%

the word "one" followed by a preposition or comma

phrases indicating sequences or repetitions of actions

No Comments

No Known Activations