Spaces:
Running
on
T4
Running
on
T4
File size: 7,716 Bytes
8ce4d25 be59b6e 8ce4d25 be59b6e 8ce4d25 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 |
schema pdf_page {
document pdf_page {
field id type string {
indexing: summary | index
match {
word
}
}
field url type string {
indexing: summary | index
}
field title type string {
indexing: summary | index
index: enable-bm25
match {
text
}
}
field page_number type int {
indexing: summary | attribute
}
field image type raw {
indexing: summary
}
field full_image type raw {
indexing: summary
}
field text type string {
indexing: summary | index
index: enable-bm25
match {
text
}
}
field embedding type tensor<int8>(patch{}, v[16]) {
indexing: attribute | index
attribute {
distance-metric: hamming
}
index {
hnsw {
max-links-per-node: 32
neighbors-to-explore-at-insert: 400
}
}
}
}
fieldset default {
fields: title, url, page_number, text
}
fieldset image {
fields: image
}
rank-profile default {
inputs {
query(qt) tensor<float>(querytoken{}, v[128])
}
function max_sim() {
expression {
sum(
reduce(
sum(
query(qt) * unpack_bits(attribute(embedding)) , v
),
max, patch
),
querytoken
)
}
}
function bm25_score() {
expression {
bm25(title) + bm25(text)
}
}
first-phase {
expression {
bm25_score
}
}
second-phase {
rerank-count: 10
expression {
max_sim
}
}
}
rank-profile retrieval-and-rerank {
inputs {
query(rq0) tensor<int8>(v[16])
query(rq1) tensor<int8>(v[16])
query(rq2) tensor<int8>(v[16])
query(rq3) tensor<int8>(v[16])
query(rq4) tensor<int8>(v[16])
query(rq5) tensor<int8>(v[16])
query(rq6) tensor<int8>(v[16])
query(rq7) tensor<int8>(v[16])
query(rq8) tensor<int8>(v[16])
query(rq9) tensor<int8>(v[16])
query(rq10) tensor<int8>(v[16])
query(rq11) tensor<int8>(v[16])
query(rq12) tensor<int8>(v[16])
query(rq13) tensor<int8>(v[16])
query(rq14) tensor<int8>(v[16])
query(rq15) tensor<int8>(v[16])
query(rq16) tensor<int8>(v[16])
query(rq17) tensor<int8>(v[16])
query(rq18) tensor<int8>(v[16])
query(rq19) tensor<int8>(v[16])
query(rq20) tensor<int8>(v[16])
query(rq21) tensor<int8>(v[16])
query(rq22) tensor<int8>(v[16])
query(rq23) tensor<int8>(v[16])
query(rq24) tensor<int8>(v[16])
query(rq25) tensor<int8>(v[16])
query(rq26) tensor<int8>(v[16])
query(rq27) tensor<int8>(v[16])
query(rq28) tensor<int8>(v[16])
query(rq29) tensor<int8>(v[16])
query(rq30) tensor<int8>(v[16])
query(rq31) tensor<int8>(v[16])
query(rq32) tensor<int8>(v[16])
query(rq33) tensor<int8>(v[16])
query(rq34) tensor<int8>(v[16])
query(rq35) tensor<int8>(v[16])
query(rq36) tensor<int8>(v[16])
query(rq37) tensor<int8>(v[16])
query(rq38) tensor<int8>(v[16])
query(rq39) tensor<int8>(v[16])
query(rq40) tensor<int8>(v[16])
query(rq41) tensor<int8>(v[16])
query(rq42) tensor<int8>(v[16])
query(rq43) tensor<int8>(v[16])
query(rq44) tensor<int8>(v[16])
query(rq45) tensor<int8>(v[16])
query(rq46) tensor<int8>(v[16])
query(rq47) tensor<int8>(v[16])
query(rq48) tensor<int8>(v[16])
query(rq49) tensor<int8>(v[16])
query(rq50) tensor<int8>(v[16])
query(rq51) tensor<int8>(v[16])
query(rq52) tensor<int8>(v[16])
query(rq53) tensor<int8>(v[16])
query(rq54) tensor<int8>(v[16])
query(rq55) tensor<int8>(v[16])
query(rq56) tensor<int8>(v[16])
query(rq57) tensor<int8>(v[16])
query(rq58) tensor<int8>(v[16])
query(rq59) tensor<int8>(v[16])
query(rq60) tensor<int8>(v[16])
query(rq61) tensor<int8>(v[16])
query(rq62) tensor<int8>(v[16])
query(rq63) tensor<int8>(v[16])
query(qt) tensor<float>(querytoken{}, v[128])
query(qtb) tensor<int8>(querytoken{}, v[16])
}
function max_sim() {
expression {
sum(
reduce(
sum(
query(qt) * unpack_bits(attribute(embedding)) , v
),
max, patch
),
querytoken
)
}
}
function max_sim_binary() {
expression {
sum(
reduce(
1/(1 + sum(
hamming(query(qtb), attribute(embedding)) ,v)
),
max,
patch
),
querytoken
)
}
}
first-phase {
expression {
max_sim_binary
}
}
second-phase {
rerank-count: 10
expression {
max_sim
}
}
}
} |