INDEX

Explanations

descriptions of historical buildings and landmarks

oai_token-act-pair · gpt-3.5-turbo

references to significant historical or architectural landmarks

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

Consumer

-0.84

 Lyft

-0.84

 diapers

-0.80

-0.79

 SNAP

-0.77

 testers

-0.75

 ASAP

-0.74

nesota

-0.74

heit

-0.74

Dispatch

-0.73

POSITIVE LOGITS

 monument

1.55

 cathedral

1.49

 monuments

1.47

 temple

1.38

 temples

1.34

 tomb

1.33

 palace

1.33

 towers

1.33

 statue

1.33

 memorial

1.32

Activations Density 0.583%

descriptions of historical buildings and landmarks

references to significant historical or architectural landmarks

No Comments

No Known Activations