INDEX

Explanations

positive outcomes or achievements

oai_token-act-pair · gpt-3.5-turbo

expressions of quality and value related to actions 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

anwhile

-0.84

 Strauss

-0.68

wx

-0.62

 horizont

-0.61

rought

-0.60

 makers

-0.60

ioxide

-0.60

 Creator

-0.59

urgical

-0.59

ennes

-0.59

POSITIVE LOGITS

sense

0.79

 impression

0.77

 contributions

0.72

 contribution

0.71

 dent

0.70

Ò

0.69

commit

0.68

 headlines

0.68

 strides

0.67

TEXT

0.66

Activations Density 0.116%

positive outcomes or achievements

expressions of quality and value related to actions or outcomes

No Comments

No Known Activations