INDEX

Explanations

phrases indicating performance or outcome assessment, particularly focusing on whether something is performing well or not

oai_token-act-pair · gpt-3.5-turbo

expressions of performance or effectiveness

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

adena

-0.79

ategory

-0.78

ory

-0.76

ruce

-0.72

İĭ

-0.72

hyde

-0.71

ules

-0.70

atto

-0.69

orical

-0.66

mitting

-0.65

POSITIVE LOGITS

 enough

1.07

enough

0.99

esley

0.79

 Enough

0.77

 behaved

0.77

espie

0.73

 suited

0.71

 liked

0.70

baum

0.68

 alright

0.66

Activations Density 0.031%

phrases indicating performance or outcome assessment, particularly focusing on whether something is performing well or not

expressions of performance or effectiveness

No Comments

No Known Activations