INDEX

Explanations

phrases related to news and events

oai_token-act-pair · gpt-3.5-turbo

references to specific people, events, or conditions in a narrative context

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 voluntary

-0.50

 hazard

-0.48

neg

-0.48

 violation

-0.47

 relinqu

-0.46

 savings

-0.45

 fooled

-0.45

 voluntarily

-0.45

 seizure

-0.45

 potential

-0.45

POSITIVE LOGITS

âĦ¢:

0.67

ï¸ı

0.57

Elsewhere

0.55

reddit

0.55

rawdownloadcloneembedreportprint

0.52

Í

0.52

Scroll

0.50

ta

0.50

 Flavoring

0.49

 Across

0.48

Activations Density 1.194%

phrases related to news and events

references to specific people, events, or conditions in a narrative context

No Comments

No Known Activations