INDEX

Explanations

references to the beginning or start of something, such as movies, events, or articles

oai_token-act-pair · gpt-3.5-turbo

mentions of "opening" events or occurrences

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.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

ieving

-0.87

rior

-0.78

ufact

-0.77

iever

-0.75

hovah

-0.74

lean

-0.74

heres

-0.74

rogens

-0.74

rome

-0.72

ievers

-0.71

POSITIVE LOGITS

 ceremony

1.11

 ceremonies

1.07

 weekend

0.92

 crawl

0.87

 credits

0.86

 sequence

0.84

 Weekend

0.83

 salv

0.83

 Hours

0.82

 rites

0.81

Activations Density 0.053%

references to the beginning or start of something, such as movies, events, or articles

mentions of "opening" events or occurrences

No Comments

No Known Activations