INDEX

Explanations

the word "cross" followed by various contexts or phrases

oai_token-act-pair · gpt-3.5-turbo

terms related to interdisciplinary or cross-functional collaboration

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

ICAN

-0.77

å§«

-0.75

 seiz

-0.66

nant

-0.66

ãĤ¼ãĤ¦ãĤ¹

-0.66

ikan

-0.64

rha

-0.64

acious

-0.63

 Ples

-0.63

Kut

-0.63

POSITIVE LOGITS

roads

1.40

bow

1.19

bows

1.03

cross

0.97

hair

0.94

cutting

0.92

bones

0.92

road

0.91

wal

0.89

bench

0.88

Activations Density 0.014%

the word "cross" followed by various contexts or phrases

terms related to interdisciplinary or cross-functional collaboration

No Comments

No Known Activations