INDEX

Explanations

specific dates and names from news articles or reports

oai_token-act-pair · gpt-3.5-turbo

citations and publication details in texts

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 9-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.9.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.9.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

 unic

-0.72

 refunds

-0.70

 torped

-0.66

 rightful

-0.62

 susp

-0.62

 Vall

-0.61

 trope

-0.60

 radius

-0.59

 discriminate

-0.58

 closet

-0.58

POSITIVE LOGITS

BBC

1.40

Editor

1.06

Narr

1.05

Guest

1.03

Contribut

0.98

Associated

0.94

EVA

0.93

Edited

0.93

Bloomberg

0.92

HAEL

0.89

Activations Density 0.073%

specific dates and names from news articles or reports

citations and publication details in texts

No Comments

No Known Activations