INDEX

Explanations

actions or processes related to moving or progressing

oai_token-act-pair · gpt-3.5-turbo

the word "through"

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.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

apest

-0.62

ilities

-0.59

uties

-0.54

 reason

-0.53

izer

-0.53

 Watch

-0.53

 excuse

-0.52

usual

-0.52

 Salam

-0.51

venge

-0.51

POSITIVE LOGITS

 through

3.04

 thru

2.41

through

2.34

 Through

2.15

Through

1.94

via

1.75

 trough

1.47

THR

1.16

 throughout

1.09

via

1.02

Activations Density 0.053%

actions or processes related to moving or progressing

the word "through"

No Comments

No Known Activations