INDEX

Explanations

locations or geographical references containing names of places

oai_token-act-pair · gpt-3.5-turbo

conjunctions and phrases indicating additions or connections

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

tools

-0.77

EEE

-0.76

argument

-0.73

Adapt

-0.72

software

-0.72

xual

-0.71

oglobin

-0.70

CEO

-0.70

anasia

-0.69

ifact

-0.68

POSITIVE LOGITS

 adjoining

1.13

 Glou

1.12

 Galile

1.09

 Cornwall

1.07

 Riverside

1.06

 Highlands

1.04

 Queens

1.04

 Islands

1.03

 Junction

1.03

 Sussex

1.02

Activations Density 0.158%

locations or geographical references containing names of places

conjunctions and phrases indicating additions or connections

No Comments

No Known Activations