INDEX

Explanations

words related to failed or botched events or situations

oai_token-act-pair · gpt-3.5-turbo

terms related to unsuccessful events or outcomes

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.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

istics

-0.84

utra

-0.78

Edge

-0.72

ledge

-0.72

selves

-0.71

region

-0.70

arya

-0.70

Whereas

-0.69

"}],"

-0.68

Center

-0.67

POSITIVE LOGITS

 attempts

0.89

 attempt

0.89

 pregnancies

0.82

 pregnancy

0.77

 miser

0.76

 abort

0.73

 failures

0.72

 Attempt

0.72

 Failed

0.71

ulously

0.71

Activations Density 0.049%

words related to failed or botched events or situations

terms related to unsuccessful events or outcomes

No Comments

No Known Activations