INDEX

Explanations

phrases indicating disapproval or criticism

oai_token-act-pair · gpt-3.5-turbo

negations or expressions of prohibition

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

 satisfactory

-0.64

 Corona

-0.64

Wid

-0.63

 irresistible

-0.62

FW

-0.61

 Hungry

-0.61

 restoration

-0.61

 enhancements

-0.60

livion

-0.60

 Panda

-0.57

POSITIVE LOGITS

 underestimated

0.77

erest

0.77

inventoryQuantity

0.75

arent

0.73

reated

0.73

 underestimate

0.72

aval

0.70

roach

0.69

 blindly

0.69

 worry

0.69

Activations Density 0.093%

phrases indicating disapproval or criticism

negations or expressions of prohibition

No Comments

No Known Activations