INDEX

Explanations

mentions of the word "mix."

oai_token-act-pair · gpt-3.5-turbo

references to mixtures or combinations of elements

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

 reper

-0.66

hood

-0.64

 Memorial

-0.64

 Absent

-0.63

agnar

-0.63

 Passage

-0.62

abad

-0.61

ASED

-0.61

 Lumpur

-0.60

ZI

-0.60

POSITIVE LOGITS

mix

1.02

Mix

0.92

 mixes

0.86

lr

0.81

mix

0.81

ers

0.81

ups

0.80

aminer

0.80

xual

0.80

ters

0.79

Activations Density 0.011%

mentions of the word "mix."

references to mixtures or combinations of elements

No Comments

No Known Activations