INDEX

Explanations

words with "ha" in them, potentially associated with laughter or joy

oai_token-act-pair · gpt-3.5-turbo

repeated instances of the word "ha," suggesting a focus on expressions of laughter or amusement

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.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

ython

-0.82

 Staples

-0.75

arks

-0.73

SPONSORED

-0.70

onite

-0.69

 Principle

-0.68

itol

-0.68

 Medline

-0.66

tsky

-0.65

 Sacrament

-0.65

POSITIVE LOGITS

ha

3.58

Ha

1.68

Ha

1.51

HA

1.49

har

1.13

 hail

1.12

1.09

ha

1.06

ho

1.06

 Hare

1.05

Activations Density 0.027%

words with "ha" in them, potentially associated with laughter or joy

repeated instances of the word "ha," suggesting a focus on expressions of laughter or amusement

No Comments

No Known Activations