INDEX

Explanations

references to giving or receiving advice

oai_token-act-pair · gpt-3.5-turbo

references to personal experiences and opinions expressed by the speaker

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 11-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.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.11.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 outdoors

-0.71

 Gron

-0.66

 indoors

-0.65

 specialization

-0.64

 havoc

-0.63

 restrictions

-0.60

 magnification

-0.60

Role

-0.59

hunt

-0.59

 inhibition

-0.58

POSITIVE LOGITS

 received

0.98

 uttered

0.95

oled

0.92

received

0.89

 gave

0.88

 receive

0.87

å¾

0.86

aped

0.86

 levied

0.84

 issued

0.82

Activations Density 0.165%

references to giving or receiving advice

references to personal experiences and opinions expressed by the speaker

No Comments

No Known Activations