INDEX

Explanations

comparative relationships between two concepts, where one concept is usually favorable or advantageous over the other

oai_token-act-pair · gpt-3.5-turbo

comparative phrases that quantify improvement or decline

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

aws

-0.73

sr

-0.73

ittees

-0.71

tags

-0.71

nar

-0.68

brew

-0.67

ptives

-0.66

hack

-0.65

kus

-0.65

stan

-0.63

POSITIVE LOGITS

uliffe

0.72

 likely

0.72

 Farage

0.68

 actu

0.67

 healthier

0.66

 likelihood

0.66

 incentive

0.66

 payoff

0.65

 flux

0.63

cyclopedia

0.63

Activations Density 0.053%

comparative relationships between two concepts, where one concept is usually favorable or advantageous over the other

comparative phrases that quantify improvement or decline

No Comments

No Known Activations