INDEX

Explanations

mentions of ongoing or completed operations or investigations

oai_token-act-pair · gpt-3.5-turbo

phrases related to operations and investigations

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.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

 Incarnation

-0.70

 hates

-0.68

 hating

-0.67

 behaving

-0.67

 Keys

-0.65

 Absolute

-0.64

 clinging

-0.62

 Monstrous

-0.62

 representations

-0.61

 liking

-0.60

POSITIVE LOGITS

 involve

1.47

 culmin

1.26

 entail

1.24

 involves

1.16

 culminated

1.15

 occur

1.13

 resulted

1.11

 commence

1.11

 lasted

1.09

 succeed

1.09

Activations Density 0.343%

mentions of ongoing or completed operations or investigations

phrases related to operations and investigations

No Comments

No Known Activations