INDEX

Explanations

dates and time-related information

oai_token-act-pair · gpt-3.5-turbo

references to events and statistics

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 ..."

-0.60

Ħ¢

-0.57

obyl

-0.55

tackle

-0.55

ById

-0.54

SPONSORED

-0.54

brain

-0.54

gone

-0.53

undo

-0.52

 extraord

-0.52

POSITIVE LOGITS

 however

0.70

 organisers

0.64

 although

0.63

ATURES

0.61

 Herrera

0.61

 owing

0.60

 Firstly

0.60

 depending

0.60

 Participants

0.59

 contestants

0.59

Activations Density 1.007%

dates and time-related information

references to events and statistics

No Comments

No Known Activations