INDEX

Explanations

unintelligible characters and potentially unrelated snippets of text

oai_token-act-pair · gpt-3.5-turbo

special characters or symbols often found in technical or coded content

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 6-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.6.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.6.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

milo

-0.87

 mathemat

-0.86

 Seym

-0.84

wana

-0.80

 hitch

-0.75

enhagen

-0.74

é¾įå¥ĳå£«

-0.73

Pwr

-0.72

ammy

-0.72

conservancy

-0.72

POSITIVE LOGITS

âĸ¬âĸ¬

1.31

âĸ¬

0.93

âĶĢâĶĢâĶĢâĶĢ

0.91

¬

0.84

ward

0.80

Â¯Â¯Â¯Â¯Â¯Â¯Â¯Â¯Â¯Â¯Â¯Â¯Â¯Â¯Â¯Â¯

0.79

hibition

0.75

vre

0.75

jah

0.75

forth

0.74

Activations Density 0.012%

unintelligible characters and potentially unrelated snippets of text

special characters or symbols often found in technical or coded content

No Comments

No Known Activations