INDEX

Explanations

words related to actions or events

oai_token-act-pair · gpt-3.5-turbo

instances of the word "took."

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

eers

-0.71

 david

-0.66

egg

-0.65

ibel

-0.61

ledge

-0.60

 ---------

-0.59

 bombardment

-0.59

 Smile

-0.58

 Trop

-0.58

 collapsing

-0.58

POSITIVE LOGITS

FINE

0.92

 Mehran

0.88

arnaev

0.87

osate

0.85

aways

0.84

inka

0.81

raltar

0.81

autions

0.79

ãĤ¤ãĥĪ

0.79

staking

0.79

Activations Density 0.034%

words related to actions or events

instances of the word "took."

No Comments

No Known Activations