INDEX

Explanations

Names of people or entities

oai_token-act-pair · gpt-3.5-turbo

proper nouns and specific names

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

ctors

-0.69

SIGN

-0.68

Spl

-0.67

ctor

-0.64

aneously

-0.64

lance

-0.61

aji

-0.60

CPC

-0.60

 presidents

-0.60

 denomin

-0.59

POSITIVE LOGITS

mith

1.35

aurus

1.30

pace

1.22

peed

1.21

hirt

1.18

hips

1.16

ource

1.14

ilver

1.10

1.09

chool

1.09

Activations Density 0.132%

Names of people or entities

proper nouns and specific names

No Comments

No Known Activations