INDEX

Explanations

scientific or factual information mentioned in different contexts

oai_token-act-pair · gpt-3.5-turbo

phrases that indicate evidence and claims related to investigations

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

 Achieve

-0.83

Effective

-0.76

Serv

-0.72

agements

-0.72

ourses

-0.67

Freedom

-0.66

 Cooldown

-0.66

externalActionCode

-0.65

oliberal

-0.65

rawdownloadcloneembedreportprint

-0.64

POSITIVE LOGITS

 circumst

1.52

evidence

1.50

 corrobor

1.49

Evidence

1.43

 evidence

1.40

 suspicions

1.39

 conjecture

1.39

 theories

1.33

 hypothesis

1.32

 suspicion

1.31

Activations Density 1.275%

scientific or factual information mentioned in different contexts

phrases that indicate evidence and claims related to investigations

No Comments

No Known Activations