INDEX

Explanations

strong opinions or commands

oai_token-act-pair · gpt-3.5-turbo

phrases that suggest action or demand consequences

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.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

é¾įå¥ĳå£«

-0.81

ģĸ

-0.65

Posts

-0.64

Ļ½

-0.64

SPA

-0.64

 Dhabi

-0.61

ECH

-0.61

Link

-0.60

boa

-0.60

sets

-0.59

POSITIVE LOGITS

 themselves

1.11

 collectively

0.88

 Rohing

0.77

selves

0.72

 respective

0.72

 uniformly

0.68

 necks

0.67

 respectively

0.66

 individually

0.66

umm

0.65

Activations Density 1.263%

strong opinions or commands

phrases that suggest action or demand consequences

No Comments

No Known Activations