INDEX

Explanations

mentions of events, actions, or decisions that did not succeed

oai_token-act-pair · gpt-3.5-turbo

instances of the word "failed."

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

enfranch

-0.78

selves

-0.70

order

-0.69

til

-0.69

Cloud

-0.67

tip

-0.66

gon

-0.66

agon

-0.66

Forward

-0.65

orthy

-0.65

POSITIVE LOGITS

 miser

1.22

 fail

0.94

 failures

0.90

 failure

0.87

fail

0.85

 fails

0.85

 horribly

0.85

 failed

0.83

 attempt

0.83

 failing

0.80

Activations Density 0.011%

mentions of events, actions, or decisions that did not succeed

instances of the word "failed."

No Comments

No Known Activations