INDEX

Explanations

words related to negative emotions, particularly grief

oai_token-act-pair · gpt-3.5-turbo

mentions of grief and related emotional distress

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

advertisement

-0.75

Foot

-0.71

oral

-0.69

UP

-0.69

ribe

-0.68

ells

-0.65

KEY

-0.65

illin

-0.65

OPA

-0.65

onomy

-0.64

POSITIVE LOGITS

 grief

1.09

 grieving

0.93

 stricken

0.91

 bere

0.87

emort

0.86

terness

0.84

 sorrow

0.81

 redes

0.80

swer

0.79

 forgive

0.77

Activations Density 0.007%

words related to negative emotions, particularly grief

mentions of grief and related emotional distress

No Comments

No Known Activations