INDEX

Explanations

literal interpretations or expressions

oai_token-act-pair · gpt-3.5-turbo

references to literal and metaphorical concepts

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

ulates

-0.71

Crash

-0.67

ramid

-0.67

arma

-0.65

shows

-0.64

 Features

-0.64

 Share

-0.64

 Privacy

-0.63

spr

-0.63

company

-0.62

POSITIVE LOGITS

 literal

1.16

orical

0.82

 analogue

0.80

 canonical

0.73

 interpretation

0.72

TY

0.71

 brunt

0.70

ãĤ¦ãĤ¹

0.70

 incarnation

0.70

alogue

0.69

Activations Density 0.005%

literal interpretations or expressions

references to literal and metaphorical concepts

No Comments

No Known Activations