INDEX

Explanations

words related to physical locations, particularly being in a specific place or country

oai_token-act-pair · gpt-3.5-turbo

instances of the phrases “here in” and “at” referencing locations

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

imon

-0.77

-0.74

————————

-0.69

emo

-0.69

0000000000000000

-0.65

Final

-0.63

ãĥİ

-0.62

Plugin

-0.62

 sshd

-0.62

Interest

-0.62

POSITIVE LOGITS

swick

0.80

order

0.77

asin

0.75

idth

0.69

ruciating

0.69

odon

0.69

isphere

0.67

erers

0.67

ibia

0.65

rha

0.64

Activations Density 0.071%

words related to physical locations, particularly being in a specific place or country

instances of the phrases “here in” and “at” referencing locations

No Comments

No Known Activations