INDEX

Explanations

words related to errors or failures

oai_token-act-pair · gpt-3.5-turbo

references to mistakes and errors

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

wired

-0.74

hibit

-0.70

rer

-0.69

rix

-0.69

cend

-0.68

 fitted

-0.65

lain

-0.65

skin

-0.65

icum

-0.63

 Supp

-0.63

POSITIVE LOGITS

 mistakes

3.86

 errors

2.59

 mistake

2.05

 flaws

1.93

 faults

1.92

 failings

1.89

errors

1.88

 Errors

1.86

 shortcomings

1.83

 failures

1.83

Activations Density 0.027%

words related to errors or failures

references to mistakes and errors

No Comments

No Known Activations