INDEX

Explanations

information related to statistics, reports, and findings mentioned in various contexts

oai_token-act-pair · gpt-3.5-turbo

references to authoritative sources or citations in the text

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

tracks

-0.76

marine

-0.76

worms

-0.76

 advoc

-0.74

thumbnails

-0.71

omorph

-0.67

RNA

-0.67

©¶æ¥µ

-0.67

boxes

-0.66

estyles

-0.66

POSITIVE LOGITS

 Franch

0.75

 Dame

0.71

 Sense

0.70

ilet

0.69

 Bunker

0.67

 Huss

0.67

 Amen

0.67

wards

0.66

 Pascal

0.64

 Presbyterian

0.64

Activations Density 0.351%

information related to statistics, reports, and findings mentioned in various contexts

references to authoritative sources or citations in the text

No Comments

No Known Activations