INDEX

Explanations

phrases indicating a source is mentioned or quoted

oai_token-act-pair · gpt-3.5-turbo

instances of the word "said."

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.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

à¦

-0.84

agra

-0.72

phal

-0.72

otion

-0.71

omore

-0.69

mol

-0.68

VIDEO

-0.68

ptives

-0.68

²¾

-0.68

ooter

-0.67

POSITIVE LOGITS

 afterward

0.77

 bluntly

0.71

doms

0.69

 analysts

0.69

 anecd

0.68

 goodbye

0.67

 sarcast

0.67

 spokeswoman

0.65

 afterwards

0.65

 brisk

0.64

Activations Density 0.249%

phrases indicating a source is mentioned or quoted

instances of the word "said."

No Comments

No Known Activations