INDEX

Explanations

adjectives indicating strong opinions or comparisons

oai_token-act-pair · gpt-3.5-turbo

adjectives and their intensifiers that describe quality or effectiveness

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

emen

-0.82

 Rooms

-0.73

etts

-0.72

othal

-0.69

 ..............

-0.69

CHAT

-0.69

 Cellular

-0.68

 Pump

-0.67

Plex

-0.65

uctions

-0.65

POSITIVE LOGITS

 tong

0.71

AS

0.70

 than

0.67

idious

0.65

Osc

0.63

iated

0.61

as

0.61

 athlet

0.59

 tract

0.58

 Than

0.58

Activations Density 0.105%

adjectives indicating strong opinions or comparisons

adjectives and their intensifiers that describe quality or effectiveness

No Comments

No Known Activations