INDEX

Explanations

words related to different types of professions, especially those that end with "-ers"

oai_token-act-pair · gpt-3.5-turbo

suffixes referring to groups or categories of people, particularly those ending in "ers"

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 forfeiture

-0.71

URRENT

-0.61

ADRA

-0.60

 conviction

-0.60

 Downing

-0.59

itars

-0.58

PUT

-0.57

 Guinea

-0.57

 numerical

-0.57

SIGN

-0.57

POSITIVE LOGITS

mith

1.24

paces

1.15

pace

1.12

linger

1.07

ucker

1.07

poon

1.06

cream

1.06

hip

1.04

haw

1.03

erker

0.98

Activations Density 0.044%

words related to different types of professions, especially those that end with "-ers"

suffixes referring to groups or categories of people, particularly those ending in "ers"

No Comments

No Known Activations