INDEX

Explanations

words related to satisfaction or displeasure

oai_token-act-pair · gpt-3.5-turbo

variations of the word "ple."

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

 Kenobi

-0.81

DERR

-0.75

ERA

-0.71

 Colonial

-0.64

 srfAttach

-0.63

eworld

-0.62

clinton

-0.62

 Ernst

-0.62

Reviewer

-0.62

rolet

-0.62

POSITIVE LOGITS

teness

1.26

mented

1.02

ple

0.97

plet

0.97

ments

0.93

asure

0.84

asures

0.81

thora

0.81

xt

0.80

asing

0.75

Activations Density 0.017%

words related to satisfaction or displeasure

variations of the word "ple."

No Comments

No Known Activations