INDEX

Explanations

words related to specific places or locations

oai_token-act-pair · gpt-3.5-turbo

occurrences of the substring "ins" within words

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

ploma

-0.64

 Governments

-0.63

 Prime

-0.62

lling

-0.62

 Achieve

-0.60

 Sigma

-0.60

lled

-0.59

 DISTR

-0.58

LIA

-0.58

llan

-0.57

POSITIVE LOGITS

pection

1.34

urance

1.17

piration

1.00

anity

0.99

ufficient

0.98

poon

0.98

piring

0.97

pect

0.96

kaya

0.95

creen

0.93

Activations Density 0.031%

words related to specific places or locations

occurrences of the substring "ins" within words

No Comments

No Known Activations