INDEX

Explanations

words with prefixes such as 're-' and 're-' words themselves

oai_token-act-pair · gpt-3.5-turbo

prefixes or variations of the prefix "re-"

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 noble

-0.71

 Relief

-0.70

 Tant

-0.70

 Slater

-0.70

 vast

-0.68

CPR

-0.68

 percentile

-0.66

 famine

-0.65

 dense

-0.65

 Slim

-0.65

POSITIVE LOGITS

introdu

1.59

elected

1.39

establish

1.39

education

1.37

examination

1.37

ordering

1.37

election

1.33

issue

1.33

release

1.32

acqu

1.32

Activations Density 0.025%

words with prefixes such as 're-' and 're-' words themselves

prefixes or variations of the prefix "re-"

No Comments

No Known Activations