INDEX

Explanations

imperatives instructing or advising actions

oai_token-act-pair · gpt-3.5-turbo

negative imperatives or cautionary advice

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

wrapper

-0.61

language

-0.61

 Reborn

-0.60

 Reloaded

-0.59

 ancest

-0.59

 Houses

-0.58

 Greenberg

-0.58

strength

-0.57

 resid

-0.57

ELD

-0.57

POSITIVE LOGITS

 bother

1.07

 hesitate

1.06

 forget

1.06

 underestimate

0.90

 worry

0.87

 discriminate

0.86

 expect

0.85

Í

0.83

erest

0.81

intend

0.80

Activations Density 0.058%

imperatives instructing or advising actions

negative imperatives or cautionary advice

No Comments

No Known Activations