INDEX

Explanations

phrases related to steps or actions needed to achieve a specific goal

oai_token-act-pair · gpt-3.5-turbo

phrases related to instructions or guidelines for achieving tasks

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

 outweigh

-0.78

 outwe

-0.68

 drowned

-0.67

marine

-0.65

 benches

-0.65

 alive

-0.63

burning

-0.63

 quot

-0.62

pez

-0.62

 harb

-0.61

POSITIVE LOGITS

aucus

0.73

itialized

0.68

 nutshell

0.66

 Brief

0.65

ertain

0.64

 recap

0.64

zbek

0.64

 meantime

0.63

 disclaimer

0.62

kinson

0.62

Activations Density 0.121%

phrases related to steps or actions needed to achieve a specific goal

phrases related to instructions or guidelines for achieving tasks

No Comments

No Known Activations