INDEX

Explanations

comparisons between two different options

oai_token-act-pair · gpt-3.5-turbo

references to indefinite articles and numerical ratings, indicating a focus on categorizing or questioning items

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.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

 Accounts

-0.74

 Cases

-0.69

 Excellent

-0.66

Events

-0.66

 arrangements

-0.65

rates

-0.64

Easy

-0.64

Reports

-0.61

Sources

-0.61

About

-0.61

POSITIVE LOGITS

 bunch

0.94

rouse

0.92

venge

0.85

 thief

0.77

 competitor

0.76

 murderer

0.76

 piece

0.76

lot

0.76

cknowled

0.74

 breaker

0.74

Activations Density 0.221%

comparisons between two different options

references to indefinite articles and numerical ratings, indicating a focus on categorizing or questioning items

No Comments

No Known Activations