INDEX

Explanations

long and specific character strings, potentially indicating coding snippets or formatting patterns

oai_token-act-pair · gpt-3.5-turbo

ellipsis or punctuation used to indicate continuation or omission

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 6-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.6.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.6.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

agra

-0.60

zar

-0.55

ãĥ´

-0.54

 Danish

-0.54

 federation

-0.54

 exha

-0.54

veland

-0.54

eka

-0.54

 envelope

-0.54

tri

-0.53

POSITIVE LOGITS

there

0.97

they

0.77

>)

0.75

say

0.73

things

0.73

then

0.72

does

0.71

you

0.68

————

0.68

âĸº

0.67

Activations Density 0.359%

long and specific character strings, potentially indicating coding snippets or formatting patterns

ellipsis or punctuation used to indicate continuation or omission

No Comments

No Known Activations