INDEX

Explanations

specific patterns or combinations of letters that don't form common words or phrases, likely related to a specialized technical field or code

oai_token-act-pair · gpt-3.5-turbo

specific descriptors and classifications related to colors and attributes

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.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

 interviewer

-0.62

 curfew

-0.60

 Kejriwal

-0.59

ĸļ

-0.58

 pardon

-0.56

 lockout

-0.55

 vigil

-0.54

 Sale

-0.54

hift

-0.52

 Roundup

-0.52

POSITIVE LOGITS

types

1.19

objects

1.19

acters

1.15

otypes

1.15

pieces

1.14

items

1.13

 equivalents

1.11

isations

1.10

stones

1.09

iences

1.09

Activations Density 0.922%

specific patterns or combinations of letters that don't form common words or phrases, likely related to a specialized technical field or code

specific descriptors and classifications related to colors and attributes

No Comments

No Known Activations