INDEX

Explanations

words related to a specific tragedy or disaster event

oai_token-act-pair · gpt-3.5-turbo

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.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

SourceFile

-0.75

 ãĤµãĥ¼ãĥĨãĤ£ãĥ¯ãĥ³

-0.72

forth

-0.69

 Wonderland

-0.66

liest

-0.61

 Dominion

-0.59

rawdownloadcloneembedreportprint

-0.58

 Predator

-0.58

 confidentiality

-0.57

 Lanc

-0.56

POSITIVE LOGITS

agin

0.78

oslav

0.73

kov

0.68

ugi

0.68

rir

0.67

Î½

0.67

icky

0.67

anka

0.67

atile

0.66

kees

0.66

Activations Density 0.084%

words related to a specific tragedy or disaster event

No Comments

No Known Activations