INDEX

Explanations

references to alphanumeric sequences with an 's' at the end, possibly related to measurement units or codes

oai_token-act-pair · gpt-3.5-turbo

occurrences of the letter 's' or its variations

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

 seas

-0.64

Material

-0.60

wise

-0.59

////////

-0.59

 sorcery

-0.58

à¤

-0.57

 reven

-0.57

pad

-0.56

 dehydration

-0.56

Dod

-0.56

POSITIVE LOGITS

ources

1.17

ouls

1.03

wana

1.01

aturated

0.98

ourced

0.98

ubs

0.93

uggest

0.93

ucker

0.93

linger

0.91

niper

0.90

Activations Density 0.184%

references to alphanumeric sequences with an 's' at the end, possibly related to measurement units or codes

occurrences of the letter 's' or its variations

No Comments

No Known Activations