INDEX

Explanations

locations in the Middle East, particularly cities and towns

oai_token-act-pair · gpt-3.5-turbo

geographical locations and associated terms in a conflict context

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.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

wcsstore

-1.15

Scot

-0.92

 FANTASY

-0.82

natureconservancy

-0.80

Lear

-0.79

Colorado

-0.76

utherford

-0.73

 Preferred

-0.72

 Rider

-0.72

OPA

-0.72

POSITIVE LOGITS

 Homs

1.08

 Idlib

1.06

 Tripoli

0.95

 mosques

0.94

 Aleppo

0.93

Kaf

0.92

 Syri

0.91

 Donetsk

0.91

 evacuated

0.90

 detonated

0.89

Activations Density 0.174%

locations in the Middle East, particularly cities and towns

geographical locations and associated terms in a conflict context

No Comments

No Known Activations