INDEX

Explanations

names of specific geographical locations

oai_token-act-pair · gpt-3.5-turbo

short sequences of letters or characters that appear frequently

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 7-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.7.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.7.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

 mosqu

-0.70

 presidents

-0.66

 longitudinal

-0.64

 capsules

-0.62

 quake

-0.62

 starters

-0.60

dec

-0.60

 psychiat

-0.60

VICE

-0.58

 capsule

-0.58

POSITIVE LOGITS

awi

0.99

inn

0.93

uddin

0.93

oola

0.91

aj

0.90

ava

0.90

urd

0.89

idd

0.89

alf

0.88

ool

0.88

Activations Density 0.135%

names of specific geographical locations

short sequences of letters or characters that appear frequently

No Comments

No Known Activations