INDEX

Explanations

web addresses

oai_token-act-pair · gpt-3.5-turbo

instances of the word "Read" followed by numerical values, indicating calls to action or prompts for further reading

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

»Ĵ

-0.70

ascal

-0.68

unta

-0.67

IDS

-0.66

avement

-0.64

bably

-0.63

oted

-0.62

AFTA

-0.61

ilogy

-0.61

 towed

-0.60

POSITIVE LOGITS

 aloud

0.90

ership

0.86

dress

0.83

Write

0.83

andowski

0.82

ers

0.79

ings

0.74

just

0.74

iness

0.74

çīĪ

0.73

Activations Density 0.027%

web addresses

instances of the word "Read" followed by numerical values, indicating calls to action or prompts for further reading

No Comments

No Known Activations