INDEX

Explanations

words related to leadership positions or hierarchies

oai_token-act-pair · gpt-3.5-turbo

terms related to competitive matchups and rankings

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

ria

-0.68

Deploy

-0.66

ãĥīãĥ©ãĤ´ãĥ³

-0.64

Ö¼

-0.63

é¾

-0.62

åº

-0.61

TEXTURE

-0.61

ãĤ¢ãĥ«

-0.60

igue

-0.60

 Cumm

-0.59

POSITIVE LOGITS

 Winston

0.69

ggle

0.66

sed

0.64

ĸļ

0.64

ikuman

0.63

pex

0.63

ivist

0.62

 Gork

0.61

 negotiator

0.60

 scissors

0.57

Activations Density 0.103%

words related to leadership positions or hierarchies

terms related to competitive matchups and rankings

No Comments

No Known Activations