INDEX

Explanations

countries or regions

oai_token-act-pair · gpt-3.5-turbo

names of states and countries

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

Filename

-0.64

RIS

-0.62

Gaw

-0.62

multipl

-0.61

-0.60

lopp

-0.60

 Niet

-0.57

adobe

-0.57

 referen

-0.57

undown

-0.55

POSITIVE LOGITS

 alike

1.41

 respectively

1.23

 versa

0.96

avia

0.76

 thereof

0.70

bia

0.68

 coasts

0.65

.>>

0.64

aucuses

0.63

 cohorts

0.62

Activations Density 0.257%

countries or regions

names of states and countries

No Comments

No Known Activations