INDEX

Explanations

adjectives describing difficulty or challenge

oai_token-act-pair · gpt-3.5-turbo

phrases that refer to difficult or controversial topics

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

wives

-0.68

 sleeps

-0.66

 shoots

-0.66

ahs

-0.66

Initialized

-0.64

dup

-0.62

leep

-0.62

 pledges

-0.60

 lashes

-0.60

 queens

-0.60

POSITIVE LOGITS

 undeniable

0.74

 attest

0.73

 understandable

0.71

 emblem

0.71

 testament

0.71

 reminder

0.70

 indicative

0.68

 evident

0.67

ICLE

0.67

¿½

0.67

Activations Density 0.482%

adjectives describing difficulty or challenge

phrases that refer to difficult or controversial topics

No Comments

No Known Activations