INDEX

Explanations

words related to emotional distress, particularly grief

oai_token-act-pair · gpt-3.5-turbo

expressions related to grief and loss

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 4-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.4.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.4.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

utable

-0.79

stay

-0.78

onomy

-0.69

akedown

-0.67

NRS

-0.65

Notable

-0.64

ioch

-0.60

POL

-0.60

uner

-0.59

TRY

-0.59

POSITIVE LOGITS

 stricken

1.05

 grief

0.87

rieved

0.78

 grieving

0.77

 sorrow

0.75

rieve

0.75

 bere

0.73

 overcome

0.73

rette

0.72

fully

0.70

Activations Density 0.015%

words related to emotional distress, particularly grief

expressions related to grief and loss

No Comments

No Known Activations