INDEX

Explanations

locations, specifically place names

oai_token-act-pair · gpt-3.5-turbo

names of geographic locations and political entities

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

ãĤ´ãĥ³

-0.90

ĸļ

-0.88

Phys

-0.86

 Grateful

-0.81

 Cthulhu

-0.74

 Wally

-0.73

 Blizz

-0.72

 Hitch

-0.72

 Lovecraft

-0.72

STEM

-0.72

POSITIVE LOGITS

atan

0.90

ifa

0.90

 regime

0.86

 Hague

0.83

 Islamist

0.81

 Sudan

0.80

 dictatorship

0.80

 Hussein

0.79

 nationalist

0.78

 assassinated

0.77

Activations Density 0.356%

locations, specifically place names

names of geographic locations and political entities

No Comments

No Known Activations