INDEX

Explanations

statements made by or about individuals in a legal or investigatory context

oai_token-act-pair · gpt-3.5-turbo

statements related to claims and admissions in legal contexts

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 11-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.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.11.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

pmwiki

-0.76

tan

-0.67

 unison

-0.65

wagon

-0.64

Pont

-0.64

Quantity

-0.63

\.

-0.63

bang

-0.61

Frameworks

-0.59

uds

-0.59

POSITIVE LOGITS

he

1.27

his

1.02

 himself

0.98

 afterward

0.92

 remorse

0.84

 later

0.78

 afterwards

0.77

she

0.75

He

0.74

 investigators

0.71

Activations Density 0.285%

statements made by or about individuals in a legal or investigatory context

statements related to claims and admissions in legal contexts

No Comments

No Known Activations