INDEX

Explanations

phrases indicating a choice or action contrary to an expected or suggested course of action

oai_token-act-pair · gpt-3.5-turbo

phrases that indicate contrasts or alternatives

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 9-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.9.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.9.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

meric

-0.80

fifth

-0.71

ancer

-0.69

izer

-0.69

lass

-0.68

most

-0.65

third

-0.65

iple

-0.65

read

-0.65

mma

-0.64

POSITIVE LOGITS

 focusing

1.12

 relying

1.10

 wasting

1.06

 letting

1.04

 blaming

0.98

 being

0.97

 acknowledging

0.97

 rever

0.96

 apologizing

0.95

 having

0.94

Activations Density 0.050%

phrases indicating a choice or action contrary to an expected or suggested course of action

phrases that indicate contrasts or alternatives

No Comments

No Known Activations