INDEX

Explanations

instances of the word "fail" with a high level of activation

oai_token-act-pair · gpt-3.5-turbo

contexts related to failure or the concept of failing

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.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

 Crescent

-0.63

ours

-0.63

 Sands

-0.61

orney

-0.61

oust

-0.60

rams

-0.60

our

-0.60

 Sterling

-0.60

arm

-0.57

 Adrian

-0.57

POSITIVE LOGITS

 fail

3.75

fail

2.41

 fails

2.37

Fail

2.36

 Fail

2.34

 failure

2.02

 succeed

1.85

 failed

1.84

 failing

1.80

Failure

1.78

Activations Density 0.013%

instances of the word "fail" with a high level of activation

contexts related to failure or the concept of failing

No Comments

No Known Activations