INDEX

Explanations

phrases indicating a specific purpose or intention

oai_token-act-pair · gpt-3.5-turbo

phrases referring to concepts or ideas that should be considered or remembered

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

itte

-0.68

imer

-0.66

Js

-0.66

quer

-0.66

Reviewed

-0.65

 Tycoon

-0.64

hell

-0.61

atform

-0.61

mouth

-0.60

EStreamFrame

-0.59

POSITIVE LOGITS

 when

0.93

 WHEN

0.73

 beforehand

0.70

rals

0.69

 whenever

0.68

oldown

0.67

enance

0.67

anza

0.64

etary

0.64

lessly

0.63

Activations Density 0.036%

phrases indicating a specific purpose or intention

phrases referring to concepts or ideas that should be considered or remembered

No Comments

No Known Activations