INDEX

Explanations

proper nouns preceded by "Ag"

oai_token-act-pair · gpt-3.5-turbo

mentions of agencies or organizations, particularly those related to news reporting

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

 Dare

-0.71

éĹĺ

-0.70

 Jenner

-0.70

 jumper

-0.67

 strand

-0.62

 Cors

-0.62

 Piper

-0.61

 Sparkle

-0.61

 macro

-0.60

 sight

-0.59

POSITIVE LOGITS

reement

1.45

reements

1.41

reed

1.30

encies

1.26

atha

1.19

nostic

1.11

assi

1.08

endas

1.08

ree

1.07

ressive

1.02

Activations Density 0.017%

proper nouns preceded by "Ag"

mentions of agencies or organizations, particularly those related to news reporting

No Comments

No Known Activations