INDEX

Explanations

references to the consulting firm "McKinsey"

oai_token-act-pair · gpt-3.5-turbo

references to McKinsey and related terms

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.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

 strains

-0.69

 strain

-0.68

char

-0.65

UV

-0.65

 garlic

-0.65

 detect

-0.64

unes

-0.63

 dose

-0.63

 deity

-0.62

ansk

-0.62

POSITIVE LOGITS

 McKin

4.21

 Polk

1.07

 Hitman

1.07

 Roose

1.06

 Bain

1.06

Kin

1.01

 Pric

0.94

 Harding

0.89

Xer

0.87

 JPMorgan

0.87

Activations Density 0.020%

references to the consulting firm "McKinsey"

references to McKinsey and related terms

No Comments

No Known Activations