INDEX

Explanations

comparative statements indicating difficulty or challenge

oai_token-act-pair · gpt-3.5-turbo

instances where the word "harder" is used to describe increasing difficulty or challenges

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

Kings

-0.71

reen

-0.69

oin

-0.67

 Loft

-0.66

 Lights

-0.66

EVA

-0.66

rophe

-0.65

Rus

-0.65

 Constantine

-0.65

 Monaco

-0.65

POSITIVE LOGITS

 than

1.08

 Than

0.85

 harder

0.81

 resil

0.78

nces

0.76

 behaved

0.75

 toget

0.75

 forgiving

0.74

 risky

0.72

 scrambled

0.69

Activations Density 0.023%

comparative statements indicating difficulty or challenge

instances where the word "harder" is used to describe increasing difficulty or challenges

No Comments

No Known Activations