INDEX

Explanations

words related to providing information or clues

oai_token-act-pair · gpt-3.5-turbo

references to hints or clues in various contexts

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

Zed

-0.70

 Nationwide

-0.67

ufact

-0.66

CN

-0.66

effic

-0.65

NCT

-0.64

orld

-0.62

 NAME

-0.61

 Chatt

-0.61

Mub

-0.60

POSITIVE LOGITS

tip

1.03

tip

1.01

sters

0.99

ster

0.98

jar

0.96

haps

0.92

 toes

0.88

sy

0.87

 tips

0.82

 iceberg

0.81

Activations Density 0.027%

words related to providing information or clues

references to hints or clues in various contexts

No Comments

No Known Activations