INDEX

Explanations

mentions of specific individuals in official or professional capacities

oai_token-act-pair · gpt-3.5-turbo

items of information related to people's names and titles

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

process

-0.70

IP

-0.64

acters

-0.63

overs

-0.60

eries

-0.60

ãĥ©

-0.57

ãĥ¥

-0.56

=#

-0.56

aires

-0.55

 aggregation

-0.55

POSITIVE LOGITS

Jr

1.25

who

1.06

who

1.05

 whose

1.04

Sr

1.01

 whom

0.96

aka

0.96

whose

0.94

 meanwhile

0.93

*/(

0.81

Activations Density 0.210%

mentions of specific individuals in official or professional capacities

items of information related to people's names and titles

No Comments

No Known Activations