INDEX

Explanations

phrases related to causing disappointment or irritation

oai_token-act-pair · gpt-3.5-turbo

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 caucuses

-0.74

eki

-0.72

oscope

-0.63

 Puzzles

-0.61

 Documents

-0.57

 acad

-0.57

 chairs

-0.57

 Chains

-0.56

 passports

-0.56

 Rocks

-0.56

POSITIVE LOGITS

needed

0.79

ashtra

0.69

aunted

0.67

awaited

0.65

 sidx

0.63

itant

0.62

haul

0.62

 admire

0.62

ocked

0.61

 Reserved

0.61

Activations Density 0.131%

phrases related to causing disappointment or irritation

No Comments

No Known Activations