INDEX

Explanations

phrases indicating frequency of actions or events

oai_token-act-pair · gpt-3.5-turbo

references to frequency of occurrence or habitual actions

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 11-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.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.11.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

ees

-0.79

olas

-0.76

etics

-0.70

endants

-0.65

EEP

-0.61

ebus

-0.59

OHN

-0.59

eus

-0.58

atz

-0.57

agle

-0.57

POSITIVE LOGITS

 weekday

1.16

day

1.13

 week

1.10

 month

1.07

 fortnight

1.05

 morning

1.05

where

1.04

time

1.00

 year

1.00

 single

0.99

Activations Density 0.045%

phrases indicating frequency of actions or events

references to frequency of occurrence or habitual actions

No Comments

No Known Activations