INDEX

Explanations

phrases related to commands or directives

oai_token-act-pair · gpt-3.5-turbo

the presence of quoted statements or dialogues

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.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

 whipping

-0.65

 fabric

-0.63

 skirts

-0.62

 Vaugh

-0.62

 harness

-0.62

 reprint

-0.61

 Odin

-0.61

 folded

-0.61

ãĥ¼ãĥ«

-0.61

 railing

-0.60

POSITIVE LOGITS

yss

0.94

Cause

0.93

Mech

0.87

',

0.84

-'

0.82

cause

0.82

affer

0.79

Interstitial

0.78

Too

0.76

Marginal

0.75

Activations Density 0.040%

phrases related to commands or directives

the presence of quoted statements or dialogues

No Comments

No Known Activations