INDEX

Explanations

pronouns followed by verifiable claims or accusations

oai_token-act-pair · gpt-3.5-turbo

references to assertions and claims made by individuals or groups

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

hang

-0.64

endif

-0.63

geries

-0.59

 Forth

-0.58

asking

-0.58

 Fishing

-0.57

 intrusion

-0.56

ettlement

-0.55

ancial

-0.55

 Vine

-0.54

POSITIVE LOGITS

 deems

1.09

 believes

0.97

 deemed

0.97

 dubbed

0.96

 termed

0.95

 considers

0.94

 deem

0.94

 alleges

0.90

 blames

0.85

ãĤ¹

0.82

Activations Density 0.154%

pronouns followed by verifiable claims or accusations

references to assertions and claims made by individuals or groups

No Comments

No Known Activations