INDEX

Explanations

names or terms with the letters "inn" in them

oai_token-act-pair · gpt-3.5-turbo

references to prominent individuals and specific names

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.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

ãĤ´ãĥ³

-0.74

 è£ıè

-0.69

ħĭ

-0.66

 backlog

-0.66

 Lumpur

-0.65

ĸļ

-0.63

Õ

-0.62

 Kens

-0.61

Ļ½

-0.61

ãĥ£

-0.61

POSITIVE LOGITS

ruary

1.35

gerald

0.97

ascript

0.88

illet

0.87

acci

0.83

renheit

0.82

ancial

0.78

Ã©r

0.78

heit

0.77

owship

0.76

Activations Density 0.260%

names or terms with the letters "inn" in them

references to prominent individuals and specific names

No Comments

No Known Activations