INDEX

Explanations

proper nouns, especially names of individuals

oai_token-act-pair · gpt-3.5-turbo

names and terms associated with individuals or entities

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

ãĤ¤ãĥĪ

-0.70

CHA

-0.66

apest

-0.64

»Ĵ

-0.64

vertisement

-0.62

é¾

-0.61

ãĥīãĥ©ãĤ´ãĥ³

-0.60

 imaginary

-0.60

SIGN

-0.60

 imitation

-0.60

POSITIVE LOGITS

eper

0.74

wed

0.70

aults

0.69

ecast

0.68

Vul

0.63

fried

0.63

 Finn

0.63

abo

0.63

pard

0.63

ember

0.61

Activations Density 0.082%

proper nouns, especially names of individuals

names and terms associated with individuals or entities

No Comments

No Known Activations