INDEX

Explanations

phrases or sentences where some entity is identified as a specific person or group

oai_token-act-pair · gpt-3.5-turbo

instances of the word "identified" indicating recognition or classification

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.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

olt

-0.70

 goodbye

-0.67

prem

-0.66

rip

-0.65

touch

-0.64

uu

-0.62

 Heaven

-0.62

pow

-0.62

 heaven

-0.61

 rain

-0.61

POSITIVE LOGITS

 identified

3.38

 identifies

2.05

identified

2.00

 identify

1.93

 identifiable

1.85

 identifying

1.80

 detected

1.58

 identification

1.57

 Identified

1.56

 described

1.48

Activations Density 0.014%

phrases or sentences where some entity is identified as a specific person or group

instances of the word "identified" indicating recognition or classification

No Comments

No Known Activations