INDEX

Explanations

comparative phrases indicating an increasing degree of a specific attribute or action

oai_token-act-pair · gpt-3.5-turbo

expressions indicating something has not happened or been achieved yet

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

rities

-0.92

tein

-0.85

ufact

-0.74

regate

-0.71

itialized

-0.70

ostics

-0.69

urated

-0.69

scribe

-0.69

packs

-0.68

ocene

-0.66

POSITIVE LOGITS

 somehow

0.71

 undecided

0.65

 again

0.65

TBD

0.64

 manages

0.63

 totaling

0.63

GAN

0.62

YY

0.62

NESS

0.62

entimes

0.61

Activations Density 0.021%

comparative phrases indicating an increasing degree of a specific attribute or action

expressions indicating something has not happened or been achieved yet

No Comments

No Known Activations