INDEX

Explanations

proper nouns related to drinks and characters

oai_token-act-pair · gpt-3.5-turbo

names or references to specific entities, particularly those related to "Dra."

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

 Monarch

-0.74

 Clover

-0.71

 Butterfly

-0.71

 Tune

-0.69

hex

-0.66

 Hedge

-0.64

 Akron

-0.63

 Crane

-0.63

 Sussex

-0.63

 prom

-0.63

POSITIVE LOGITS

rys

0.94

enei

0.91

sten

0.90

izens

0.90

ildo

0.88

cffff

0.86

ven

0.85

izen

0.85

vana

0.84

thur

0.83

Activations Density 0.023%

proper nouns related to drinks and characters

names or references to specific entities, particularly those related to "Dra."

No Comments

No Known Activations