INDEX

Explanations

the word "a" in various contexts and phrases

oai_token-act-pair · gpt-3.5-turbo

repeated instances of the article "a"

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.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

 Jagu

-0.73

 quotas

-0.70

 Clarkson

-0.65

 advis

-0.65

anamo

-0.65

Keefe

-0.64

[*

-0.62

 Vaugh

-0.61

Finish

-0.60

killers

-0.60

POSITIVE LOGITS

cess

0.84

sexual

0.77

uras

0.71

 steady

0.70

ird

0.69

lder

0.69

 glimpse

0.67

rians

0.66

 couple

0.66

 guest

0.65

Activations Density 0.059%

the word "a" in various contexts and phrases

repeated instances of the article "a"

No Comments

No Known Activations