INDEX

Explanations

technical language related to actions or developments

oai_token-act-pair · gpt-3.5-turbo

references to quantities and measurements in situational contexts

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 democracies

-0.85

 appoint

-0.84

 transact

-0.83

 treaties

-0.81

 continuity

-0.77

 antitrust

-0.74

 undermine

-0.72

 annex

-0.71

 agendas

-0.70

 assass

-0.69

POSITIVE LOGITS

Redditor

1.08

 hikers

0.91

Amazing

0.90

 Guinness

0.90

 passers

0.86

PHOTOS

0.85

Magikarp

0.85

Luckily

0.81

Congratulations

0.81

Photos

0.80

Activations Density 1.135%

technical language related to actions or developments

references to quantities and measurements in situational contexts

No Comments

No Known Activations