INDEX

Explanations

phrases related to explaining or understanding a concept

oai_token-act-pair · gpt-3.5-turbo

instances that indicate understanding or knowledge about a topic

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

wives

-0.65

Sov

-0.62

 unfocusedRange

-0.61

 approvals

-0.58

Virgin

-0.57

 registry

-0.57

 tolerated

-0.56

 Votes

-0.56

 canceled

-0.55

 subcontract

-0.55

POSITIVE LOGITS

 briefly

0.90

 empir

0.84

 analogy

0.83

 examine

0.79

alogy

0.78

uncture

0.76

 illuminating

0.76

 illuminate

0.75

 analyze

0.75

 analy

0.74

Activations Density 0.573%

phrases related to explaining or understanding a concept

instances that indicate understanding or knowledge about a topic

No Comments

No Known Activations