INDEX

Explanations

phrases related to detailed investigative work

oai_token-act-pair · gpt-3.5-turbo

references to investigations or legal proceedings

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

Pwr

-0.57

asma

-0.55

cise

-0.55

.<

-0.53

ozy

-0.53

 CrossRef

-0.51

cellaneous

-0.51

!.

-0.51

olulu

-0.50

+.

-0.50

POSITIVE LOGITS

 hadn

0.75

 exists

0.72

 existed

0.72

 could

0.68

 might

0.68

 outweigh

0.67

 justifies

0.67

 wasn

0.67

 should

0.67

 would

0.66

Activations Density 1.454%

phrases related to detailed investigative work

references to investigations or legal proceedings

No Comments

No Known Activations