INDEX

Explanations

incidents or actions related to news and media

oai_token-act-pair · gpt-3.5-turbo

phrases related to updates and reports on various topics

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 4-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.4.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.4.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

sic

-0.74

 someday

-0.73

Dyn

-0.63

 bots

-0.61

 Shards

-0.60

 Alchemy

-0.59

ichick

-0.59

oteric

-0.58

 Transformers

-0.57

 deductions

-0.56

POSITIVE LOGITS

ccording

0.85

Synopsis

0.76

BBC

0.72

umbai

0.67

estern

0.66

Posted

0.64

March

0.64

SAN

0.64

haw

0.64

Activations Density 0.136%

incidents or actions related to news and media

phrases related to updates and reports on various topics

No Comments

No Known Activations