INDEX

Explanations

words related to "exposure" or "reveal"

oai_token-act-pair · gpt-3.5-turbo

instances of negations and negative prefixes in words

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

 arteries

-0.65

 inhabited

-0.64

yip

-0.63

 obstruction

-0.61

eele

-0.60

 acqu

-0.59

 manned

-0.59

 disliked

-0.59

ibrary

-0.58

 Carnegie

-0.58

POSITIVE LOGITS

cled

1.02

atorium

0.87

entials

0.84

lished

0.83

VERTISEMENT

0.82

emn

0.82

lishing

0.80

ROM

0.77

rocal

0.76

CAST

0.76

Activations Density 0.116%

words related to "exposure" or "reveal"

instances of negations and negative prefixes in words

No Comments

No Known Activations