| <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 1200 3119" width="1200" height="3119" font-size="14"> |
| <style> |
| :root { |
| --bg: #ffffff; --fg: #1b1f24; --muted: #6b7280; --panel: #f6f8fa; --grid: #e5e7eb; |
| --embed: #dbeafe; --embed-s: #3b82f6; |
| --attn: #cffafe; --attn-s: #06b6d4; |
| --mamba: #dcfce7; --mamba-s: #22c55e; |
| --linattn: #fce7f3; --linattn-s: #ec4899; |
| --recur: #ede9fe; --recur-s: #8b5cf6; |
| --moe: #ffedd5; --moe-s: #f97316; |
| --mlp: #ede9fe; --mlp-s: #8b5cf6; |
| --norm: #e5e7eb; --norm-s: #9ca3af; |
| --head: #fee2e2; --head-s: #ef4444; |
| --config: #f1f5f9; --config-s: #64748b; |
| --rope: #fef9c3; --rope-s: #eab308; |
| --layer: #f8fafc; --layer-s: #cbd5e1; |
| --io: #f1f5f9; --io-s: #94a3b8; |
| --soft: #fae8ff; --soft-s: #c026d3; |
| --add: #ffffff; --add-s: #475569; |
| --block-s: #94a3b8; |
| --residual: #f59e0b; |
| --added: #16a34a; --over: #d97706; --deleted: #dc2626; |
| --lt-full: #06b6d4; --lt-sliding: #3b82f6; --lt-chunked: #8b5cf6; |
| --lt-compressed: #f97316; --lt-heavy: #dc2626; --lt-linear: #ec4899; --lt-mamba: #22c55e; |
| --cell-on: #0ea5e9; --cell-off: #e5e7eb; |
| --vision: #dcfce7; --vision-s: #16a34a; --audio: #fae8ff; --audio-s: #c026d3; |
| --proj: #fef3c7; --proj-s: #d97706; --xattn: #db2777; |
| --conv: #d1fae5; --conv-s: #10b981; --act: #ecfccb; --act-s: #65a30d; |
| --pool: #e0f2fe; --pool-s: #0284c7; --quant: #fae8ff; --quant-s: #c026d3; |
| } |
| @media (prefers-color-scheme: dark) { |
| :root { |
| --bg: #0d1117; --fg: #e6edf3; --muted: #8b949e; --panel: #161b22; --grid: #30363d; |
| --embed: #172554; --attn: #083344; --mamba: #052e16; --linattn: #500724; |
| --recur: #2e1065; --moe: #431407; --mlp: #2e1065; --norm: #21262d; --head: #450a0a; |
| --config: #1e293b; --rope: #422006; --layer: #161b22; |
| --conv: #022c22; --act: #1a2e05; --pool: #082f49; --quant: #3b0764; --proj: #422006; |
| --io: #1e293b; --soft: #3b0764; --add: #0d1117; --cell-off: #21262d; |
| } |
| } |
| .bg { fill: var(--bg); } |
| text { font-family: ui-sans-serif, -apple-system, "Segoe UI", Roboto, sans-serif; fill: var(--fg); } |
| .title { font-size: 22px; font-weight: 700; } |
| .subtitle { font-size: 13px; fill: var(--muted); } |
| .box-label { font-size: 14px; font-weight: 600; } |
| .box-label.sm { font-size: 12.5px; } |
| .box-sub { font-size: 11px; fill: var(--muted); } |
| .glyph { font-size: 18px; font-weight: 700; fill: var(--add-s); } |
| .badge { font-size: 12px; font-weight: 700; fill: var(--fg); } |
| .facts-k { font-size: 11.5px; fill: var(--muted); } |
| .facts-v { font-size: 11.5px; font-weight: 600; } |
| .legend-t { font-size: 11.5px; fill: var(--fg); } |
| .panel { fill: var(--panel); stroke: var(--grid); } |
| rect.b { rx: 9; stroke-width: 1.6; } |
| .c-embed { fill: var(--embed); stroke: var(--embed-s); } |
| .c-attn { fill: var(--attn); stroke: var(--attn-s); } |
| .c-mamba { fill: var(--mamba); stroke: var(--mamba-s); } |
| .c-linattn { fill: var(--linattn); stroke: var(--linattn-s); } |
| .c-recur { fill: var(--recur); stroke: var(--recur-s); } |
| .c-moe { fill: var(--moe); stroke: var(--moe-s); } |
| .c-mlp { fill: var(--mlp); stroke: var(--mlp-s); } |
| .c-norm { fill: var(--norm); stroke: var(--norm-s); } |
| .c-head { fill: var(--head); stroke: var(--head-s); } |
| .c-config{ fill: var(--config);stroke: var(--config-s); } |
| .c-rope { fill: var(--rope); stroke: var(--rope-s); } |
| .c-proj { fill: var(--proj); stroke: var(--proj-s); } |
| .c-conv { fill: var(--conv); stroke: var(--conv-s); } |
| .c-act { fill: var(--act); stroke: var(--act-s); } |
| .c-pool { fill: var(--pool); stroke: var(--pool-s); } |
| .c-quant { fill: var(--quant); stroke: var(--quant-s); } |
| .c-layer { fill: var(--layer); stroke: var(--layer-s); } |
| .c-io { fill: var(--io); stroke: var(--io-s); } |
| .c-soft { fill: var(--soft); stroke: var(--soft-s); } |
| .c-add { fill: var(--add); stroke: var(--add-s); } |
| .c-block { fill: none; stroke: var(--block-s); stroke-width: 1.6; stroke-dasharray: 7 5; } |
| .c-lt-full { fill: var(--lt-full); stroke: var(--lt-full); } |
| .c-lt-sliding { fill: var(--lt-sliding); stroke: var(--lt-sliding); } |
| .c-lt-chunked { fill: var(--lt-chunked); stroke: var(--lt-chunked); } |
| .c-lt-compressed { fill: var(--lt-compressed); stroke: var(--lt-compressed); } |
| .c-lt-heavy { fill: var(--lt-heavy); stroke: var(--lt-heavy); } |
| .c-lt-linear { fill: var(--lt-linear); stroke: var(--lt-linear); } |
| .c-lt-mamba { fill: var(--lt-mamba); stroke: var(--lt-mamba); } |
| .cell-on { fill: var(--cell-on); } |
| .cell-off { fill: var(--cell-off); } |
| .grid-frame { fill: none; stroke: var(--grid); stroke-width: 1; } |
| .mask-bg { fill: var(--cell-off); } |
| .mask-on { fill: #22c55e; } |
| .mask-div { stroke: var(--fg); stroke-width: 1.5; stroke-dasharray: 3 2; } |
| .c-vision { fill: var(--vision); stroke: var(--vision-s); } |
| .c-audio { fill: var(--audio); stroke: var(--audio-s); } |
| .c-proj { fill: var(--proj); stroke: var(--proj-s); } |
| .c-sub { fill: var(--bg); stroke: var(--block-s); stroke-width: 1.2; } |
| .sec-h { font-size: 12px; font-weight: 700; } |
| .sec-hbar { fill: var(--bg); opacity: 0.82; } |
| .residual.xattn { stroke: var(--xattn); stroke-width: 2.4; } |
| .ghost { opacity: 0.32; stroke-dasharray: 4 3; } |
| .ch-added rect.b, rect.b.ch-added { stroke: var(--added); stroke-width: 3.2; } |
| .ch-over rect.b, rect.b.ch-over { stroke: var(--over); stroke-width: 3.2; } |
| .ch-deleted rect.b, rect.b.ch-deleted { stroke: var(--deleted); stroke-width: 3.2; } |
| .edge { stroke: var(--grid); stroke-width: 2; } |
| .flow { stroke: var(--grid); stroke-width: 2; fill: none; } |
| .residual { stroke: var(--residual); stroke-width: 2; fill: none; } |
| .rope { stroke: var(--rope-s); stroke-width: 2.2; fill: none; } |
| .xattn { stroke: var(--xattn); stroke-width: 2.4; fill: none; } |
| .cell-idx { font-size: 9px; fill: #ffffff; font-weight: 600; } |
| .sky { fill: #bae6fd; } .sun { fill: #fde047; } .hill { fill: #4ade80; } |
| </style> |
| <defs><marker id="ah-flow" viewBox="0 0 10 10" refX="8" refY="5" markerWidth="7" markerHeight="7" orient="auto-start-reverse"><path d="M0,0 L10,5 L0,10 z" fill="var(--grid)"/></marker><marker id="ah-residual" viewBox="0 0 10 10" refX="8" refY="5" markerWidth="7" markerHeight="7" orient="auto-start-reverse"><path d="M0,0 L10,5 L0,10 z" fill="var(--residual)"/></marker><marker id="ah-rope" viewBox="0 0 10 10" refX="8" refY="5" markerWidth="7" markerHeight="7" orient="auto-start-reverse"><path d="M0,0 L10,5 L0,10 z" fill="var(--rope-s)"/></marker><marker id="ah-xattn" viewBox="0 0 10 10" refX="8" refY="5" markerWidth="7" markerHeight="7" orient="auto-start-reverse"><path d="M0,0 L10,5 L0,10 z" fill="var(--xattn)"/></marker></defs> |
| <rect class="bg" x="0" y="0" width="1200" height="3119"/> |
| <text class="title" x="24" y="34">bark</text> |
| <text class="subtitle" x="24" y="54">BarkModel · staged data flow (structure)</text> |
| <g class=""><rect class="b c-sub " x="300" y="1480" width="580" height="1611" rx="11"/><rect class="sec-hbar" x="305" y="1484" width="570" height="18" rx="5"/><text class="sec-h" x="312" y="1497">codec_model · EncodecModel</text></g> |
| <g class=""><rect class="b c-sub " x="310" y="2225" width="560" height="748" rx="11"/><rect class="sec-hbar" x="315" y="2229" width="550" height="18" rx="5"/><text class="sec-h" x="322" y="2242">decoder · EncodecDecoder</text></g> |
| <g class=""><rect class="b c-sub " x="310" y="1508" width="560" height="710" rx="11"/><rect class="sec-hbar" x="315" y="1512" width="550" height="18" rx="5"/><text class="sec-h" x="322" y="1525">encoder · EncodecEncoder</text></g> |
| <g class=""><rect class="b c-sub " x="318" y="2250" width="544" height="710" rx="11"/><rect class="sec-hbar" x="323" y="2254" width="534" height="18" rx="5"/><text class="sec-h" x="330" y="2267">layers · ModuleList [16]</text></g> |
| <g class=""><rect class="b c-sub " x="318" y="1533" width="544" height="672" rx="11"/><rect class="sec-hbar" x="323" y="1537" width="534" height="18" rx="5"/><text class="sec-h" x="330" y="1550">layers · ModuleList [16]</text></g> |
| <g class=""><rect class="b c-sub " x="300" y="181" width="580" height="409" rx="11"/><rect class="sec-hbar" x="305" y="185" width="570" height="18" rx="5"/><text class="sec-h" x="312" y="198">semantic · SemanticModel</text></g> |
| <g class=""><rect class="b c-sub " x="300" y="614" width="580" height="409" rx="11"/><rect class="sec-hbar" x="305" y="618" width="570" height="18" rx="5"/><text class="sec-h" x="312" y="631">coarse_acoustics · CoarseModel</text></g> |
| <g class=""><rect class="b c-sub " x="300" y="1047" width="580" height="409" rx="11"/><rect class="sec-hbar" x="305" y="1051" width="570" height="18" rx="5"/><text class="sec-h" x="312" y="1064">fine_acoustics · FineModel</text></g> |
| <g class=""><rect class="b c-sub " x="326" y="2409" width="528" height="442" rx="11"/><rect class="sec-hbar" x="331" y="2413" width="518" height="18" rx="5"/><text class="sec-h" x="338" y="2426">[2:14] · 4× (ELU, EncodecConvTranspose1d, EncodecResnetBlock)</text></g> |
| <g class=""><rect class="b c-sub " x="326" y="1558" width="528" height="404" rx="11"/><rect class="sec-hbar" x="331" y="1562" width="518" height="18" rx="5"/><text class="sec-h" x="338" y="1575">[0:12] · 4× (EncodecConv1d, EncodecResnetBlock, ELU)</text></g> |
| <g class=""><rect class="b c-sub " x="310" y="296" width="560" height="252" rx="11"/><rect class="sec-hbar" x="315" y="300" width="550" height="18" rx="5"/><text class="sec-h" x="322" y="313">layers · 12× BarkBlock</text></g> |
| <g class=""><rect class="b c-sub " x="310" y="729" width="560" height="252" rx="11"/><rect class="sec-hbar" x="315" y="733" width="550" height="18" rx="5"/><text class="sec-h" x="322" y="746">layers · 12× BarkBlock</text></g> |
| <g class=""><rect class="b c-sub " x="310" y="1162" width="560" height="252" rx="11"/><rect class="sec-hbar" x="315" y="1166" width="550" height="18" rx="5"/><text class="sec-h" x="322" y="1179">layers · 12× BarkBlock</text></g> |
| <g class=""><rect class="b c-sub " x="334" y="1650" width="512" height="270" rx="11"/><rect class="sec-hbar" x="339" y="1654" width="502" height="18" rx="5"/><text class="sec-h" x="346" y="1667">1 · EncodecResnetBlock</text></g> |
| <g class=""><rect class="b c-sub " x="334" y="2568" width="512" height="270" rx="11"/><rect class="sec-hbar" x="339" y="2572" width="502" height="18" rx="5"/><text class="sec-h" x="346" y="2585">4 · EncodecResnetBlock</text></g> |
| <g class=""><rect class="b c-sub " x="342" y="1675" width="496" height="165" rx="11"/><rect class="sec-hbar" x="347" y="1679" width="486" height="18" rx="5"/><text class="sec-h" x="354" y="1692">block · ModuleList [4]</text></g> |
| <g class=""><rect class="b c-sub " x="342" y="2593" width="496" height="165" rx="11"/><rect class="sec-hbar" x="347" y="2597" width="486" height="18" rx="5"/><text class="sec-h" x="354" y="2610">block · ModuleList [4]</text></g> |
| <g class=""><rect class="b c-sub " x="350" y="1700" width="480" height="127" rx="11"/><rect class="sec-hbar" x="355" y="1704" width="470" height="18" rx="5"/><text class="sec-h" x="362" y="1717">[0:4] · 2× (ELU, EncodecConv1d)</text></g> |
| <g class=""><rect class="b c-sub " x="350" y="2618" width="480" height="127" rx="11"/><rect class="sec-hbar" x="355" y="2622" width="470" height="18" rx="5"/><text class="sec-h" x="362" y="2635">[0:4] · 2× (ELU, EncodecConv1d)</text></g> |
| <g class=""><rect class="b c-quant " x="310" y="2980" width="560" height="98" rx="11"/><rect class="sec-hbar" x="315" y="2984" width="550" height="18" rx="5"/><text class="sec-h" x="322" y="2997">quantizer · EncodecResidualVectorQuantizer</text></g> |
| <g class=""><rect class="b c-conv " x="334" y="2463" width="512" height="98" rx="11"/><rect class="sec-hbar" x="339" y="2467" width="502" height="18" rx="5"/><text class="sec-h" x="346" y="2480">3 · EncodecConvTranspose1d</text></g> |
| <g class=""><rect class="b c-sub " x="318" y="350" width="544" height="89" rx="11"/><rect class="sec-hbar" x="323" y="354" width="534" height="18" rx="5"/><text class="sec-h" x="330" y="367">attn · BarkSelfAttention</text></g> |
| <g class=""><rect class="b c-sub " x="318" y="446" width="544" height="89" rx="11"/><rect class="sec-hbar" x="323" y="450" width="534" height="18" rx="5"/><text class="sec-h" x="330" y="463">mlp · BarkMLP</text></g> |
| <g class=""><rect class="b c-sub " x="318" y="783" width="544" height="89" rx="11"/><rect class="sec-hbar" x="323" y="787" width="534" height="18" rx="5"/><text class="sec-h" x="330" y="800">attn · BarkSelfAttention</text></g> |
| <g class=""><rect class="b c-sub " x="318" y="879" width="544" height="89" rx="11"/><rect class="sec-hbar" x="323" y="883" width="534" height="18" rx="5"/><text class="sec-h" x="330" y="896">mlp · BarkMLP</text></g> |
| <g class=""><rect class="b c-sub " x="318" y="1216" width="544" height="89" rx="11"/><rect class="sec-hbar" x="323" y="1220" width="534" height="18" rx="5"/><text class="sec-h" x="330" y="1233">attn · BarkSelfAttention</text></g> |
| <g class=""><rect class="b c-sub " x="318" y="1312" width="544" height="89" rx="11"/><rect class="sec-hbar" x="323" y="1316" width="534" height="18" rx="5"/><text class="sec-h" x="330" y="1329">mlp · BarkMLP</text></g> |
| <g class=""><rect class="b c-quant " x="318" y="3005" width="544" height="60" rx="11"/><rect class="sec-hbar" x="323" y="3009" width="534" height="18" rx="5"/><text class="sec-h" x="330" y="3022">layers · 32× EncodecVectorQuantization</text></g> |
| <g class=""><rect class="b c-conv " x="326" y="1969" width="528" height="60" rx="11"/><rect class="sec-hbar" x="331" y="1973" width="518" height="18" rx="5"/><text class="sec-h" x="338" y="1986">12 · EncodecConv1d</text></g> |
| <g class=""><rect class="b c-recur " x="326" y="2036" width="528" height="60" rx="11"/><rect class="sec-hbar" x="331" y="2040" width="518" height="18" rx="5"/><text class="sec-h" x="338" y="2053">13 · EncodecLSTM</text></g> |
| <g class=""><rect class="b c-conv " x="326" y="2132" width="528" height="60" rx="11"/><rect class="sec-hbar" x="331" y="2136" width="518" height="18" rx="5"/><text class="sec-h" x="338" y="2149">15 · EncodecConv1d</text></g> |
| <g class=""><rect class="b c-conv " x="326" y="2275" width="528" height="60" rx="11"/><rect class="sec-hbar" x="331" y="2279" width="518" height="18" rx="5"/><text class="sec-h" x="338" y="2292">0 · EncodecConv1d</text></g> |
| <g class=""><rect class="b c-recur " x="326" y="2342" width="528" height="60" rx="11"/><rect class="sec-hbar" x="331" y="2346" width="518" height="18" rx="5"/><text class="sec-h" x="338" y="2359">1 · EncodecLSTM</text></g> |
| <g class=""><rect class="b c-conv " x="326" y="2887" width="528" height="60" rx="11"/><rect class="sec-hbar" x="331" y="2891" width="518" height="18" rx="5"/><text class="sec-h" x="338" y="2904">15 · EncodecConv1d</text></g> |
| <g class=""><rect class="b c-conv " x="334" y="1583" width="512" height="60" rx="11"/><rect class="sec-hbar" x="339" y="1587" width="502" height="18" rx="5"/><text class="sec-h" x="346" y="1600">0 · EncodecConv1d</text></g> |
| <g class=""><rect class="b c-conv " x="342" y="1847" width="496" height="60" rx="11"/><rect class="sec-hbar" x="347" y="1851" width="486" height="18" rx="5"/><text class="sec-h" x="354" y="1864">shortcut · EncodecConv1d</text></g> |
| <g class=""><rect class="b c-conv " x="342" y="2488" width="496" height="60" rx="11"/><rect class="sec-hbar" x="347" y="2492" width="486" height="18" rx="5"/><text class="sec-h" x="354" y="2505">conv · ParametrizedConvTranspose1d</text></g> |
| <g class=""><rect class="b c-conv " x="342" y="2765" width="496" height="60" rx="11"/><rect class="sec-hbar" x="347" y="2769" width="486" height="18" rx="5"/><text class="sec-h" x="354" y="2782">shortcut · EncodecConv1d</text></g> |
| <g class=""><rect class="b c-conv " x="358" y="1754" width="464" height="60" rx="11"/><rect class="sec-hbar" x="363" y="1758" width="454" height="18" rx="5"/><text class="sec-h" x="370" y="1771">1 · EncodecConv1d</text></g> |
| <g class=""><rect class="b c-conv " x="358" y="2672" width="464" height="60" rx="11"/><rect class="sec-hbar" x="363" y="2676" width="454" height="18" rx="5"/><text class="sec-h" x="370" y="2689">1 · EncodecConv1d</text></g> |
| <polyline class="flow" points="590.0,157 590.0,181" marker-end="url(#ah-flow)"/> |
| <polyline class="flow" points="590.0,590 590.0,614" marker-end="url(#ah-flow)"/> |
| <polyline class="flow" points="590.0,1023 590.0,1047" marker-end="url(#ah-flow)"/> |
| <polyline class="flow" points="590.0,1456 590.0,1480" marker-end="url(#ah-flow)"/> |
| <g class=""><rect class="b c-io" x="300" y="108" width="580" height="49" rx="24"/><text class="box-label sm" x="590.0" y="130.5" text-anchor="middle">inputs</text><text class="box-sub" x="590.0" y="145.5" text-anchor="middle">input_ids / pixel_values / input_values</text></g> |
| <g class=""><title>input_embeds_layer: Embedding [10048×768]</title><rect class="b c-embed" x="310" y="209" width="560" height="22" rx="9"/><text class="box-label sm" x="590.0" y="224.5" text-anchor="middle">input_embeds_layer Embedding [10048×768]</text></g> |
| <g class=""><title>position_embeds_layer: Embedding [1024×768]</title><rect class="b c-embed" x="310" y="238" width="560" height="22" rx="9"/><text class="box-label sm" x="590.0" y="253.5" text-anchor="middle">position_embeds_layer Embedding [1024×768]</text></g> |
| <g class=""><title>drop: Dropout</title><rect class="b c-sub" x="310" y="267" width="560" height="22" rx="9"/><text class="box-label sm" x="590.0" y="282.5" text-anchor="middle">drop Dropout</text></g> |
| <g class=""><title>layernorm_1: LayerNorm 768</title><rect class="b c-norm" x="318" y="321" width="268" height="22" rx="9"/><text class="box-label sm" x="452.0" y="336.5" text-anchor="middle">layernorm_1 LayerNorm 768</text></g> |
| <g class=""><title>layernorm_2: LayerNorm 768</title><rect class="b c-norm" x="593" y="321" width="268" height="22" rx="9"/><text class="box-label sm" x="727.0" y="336.5" text-anchor="middle">layernorm_2 LayerNorm 768</text></g> |
| <g class=""><title>attn_dropout: Dropout</title><rect class="b c-sub" x="326" y="375" width="171" height="22" rx="9"/><text class="box-label sm" x="411.5" y="390.5" text-anchor="middle">attn_dropout Dropout</text></g> |
| <g class=""><title>resid_dropout: Dropout</title><rect class="b c-sub" x="504" y="375" width="171" height="22" rx="9"/><text class="box-label sm" x="589.5" y="390.5" text-anchor="middle">resid_dropout Dropout</text></g> |
| <g class=""><title>att_proj: Linear [768→2304]</title><rect class="b c-proj" x="682" y="375" width="171" height="22" rx="9"/><text class="box-label sm" x="767.5" y="390.5" text-anchor="middle">att_proj [768→2304]</text></g> |
| <g class=""><title>out_proj: Linear [768→768]</title><rect class="b c-proj" x="326" y="404" width="528" height="22" rx="9"/><text class="box-label sm" x="590.0" y="419.5" text-anchor="middle">out_proj [768→768]</text></g> |
| <g class=""><title>in_proj: Linear [768→3072]</title><rect class="b c-proj" x="326" y="471" width="171" height="22" rx="9"/><text class="box-label sm" x="411.5" y="486.5" text-anchor="middle">in_proj [768→3072]</text></g> |
| <g class=""><title>out_proj: Linear [3072→768]</title><rect class="b c-proj" x="504" y="471" width="171" height="22" rx="9"/><text class="box-label sm" x="589.5" y="486.5" text-anchor="middle">out_proj [3072→768]</text></g> |
| <g class=""><title>dropout: Dropout</title><rect class="b c-sub" x="682" y="471" width="171" height="22" rx="9"/><text class="box-label sm" x="767.5" y="486.5" text-anchor="middle">dropout Dropout</text></g> |
| <g class=""><title>gelu: GELU</title><rect class="b c-act" x="326" y="500" width="528" height="22" rx="9"/><text class="box-label sm" x="590.0" y="515.5" text-anchor="middle">gelu GELU</text></g> |
| <g class=""><title>layernorm_final: LayerNorm 768</title><rect class="b c-norm" x="310" y="555" width="276" height="22" rx="9"/><text class="box-label sm" x="448.0" y="570.5" text-anchor="middle">layernorm_final LayerNorm 768</text></g> |
| <g class=""><title>lm_head: Linear [768→10048]</title><rect class="b c-head" x="593" y="555" width="276" height="22" rx="9"/><text class="box-label sm" x="731.0" y="570.5" text-anchor="middle">lm_head [768→10048]</text></g> |
| <g class=""><title>input_embeds_layer: Embedding [10048×768]</title><rect class="b c-embed" x="310" y="642" width="560" height="22" rx="9"/><text class="box-label sm" x="590.0" y="657.5" text-anchor="middle">input_embeds_layer Embedding [10048×768]</text></g> |
| <g class=""><title>position_embeds_layer: Embedding [1024×768]</title><rect class="b c-embed" x="310" y="671" width="560" height="22" rx="9"/><text class="box-label sm" x="590.0" y="686.5" text-anchor="middle">position_embeds_layer Embedding [1024×768]</text></g> |
| <g class=""><title>drop: Dropout</title><rect class="b c-sub" x="310" y="700" width="560" height="22" rx="9"/><text class="box-label sm" x="590.0" y="715.5" text-anchor="middle">drop Dropout</text></g> |
| <g class=""><title>layernorm_1: LayerNorm 768</title><rect class="b c-norm" x="318" y="754" width="268" height="22" rx="9"/><text class="box-label sm" x="452.0" y="769.5" text-anchor="middle">layernorm_1 LayerNorm 768</text></g> |
| <g class=""><title>layernorm_2: LayerNorm 768</title><rect class="b c-norm" x="593" y="754" width="268" height="22" rx="9"/><text class="box-label sm" x="727.0" y="769.5" text-anchor="middle">layernorm_2 LayerNorm 768</text></g> |
| <g class=""><title>attn_dropout: Dropout</title><rect class="b c-sub" x="326" y="808" width="171" height="22" rx="9"/><text class="box-label sm" x="411.5" y="823.5" text-anchor="middle">attn_dropout Dropout</text></g> |
| <g class=""><title>resid_dropout: Dropout</title><rect class="b c-sub" x="504" y="808" width="171" height="22" rx="9"/><text class="box-label sm" x="589.5" y="823.5" text-anchor="middle">resid_dropout Dropout</text></g> |
| <g class=""><title>att_proj: Linear [768→2304]</title><rect class="b c-proj" x="682" y="808" width="171" height="22" rx="9"/><text class="box-label sm" x="767.5" y="823.5" text-anchor="middle">att_proj [768→2304]</text></g> |
| <g class=""><title>out_proj: Linear [768→768]</title><rect class="b c-proj" x="326" y="837" width="528" height="22" rx="9"/><text class="box-label sm" x="590.0" y="852.5" text-anchor="middle">out_proj [768→768]</text></g> |
| <g class=""><title>in_proj: Linear [768→3072]</title><rect class="b c-proj" x="326" y="904" width="171" height="22" rx="9"/><text class="box-label sm" x="411.5" y="919.5" text-anchor="middle">in_proj [768→3072]</text></g> |
| <g class=""><title>out_proj: Linear [3072→768]</title><rect class="b c-proj" x="504" y="904" width="171" height="22" rx="9"/><text class="box-label sm" x="589.5" y="919.5" text-anchor="middle">out_proj [3072→768]</text></g> |
| <g class=""><title>dropout: Dropout</title><rect class="b c-sub" x="682" y="904" width="171" height="22" rx="9"/><text class="box-label sm" x="767.5" y="919.5" text-anchor="middle">dropout Dropout</text></g> |
| <g class=""><title>gelu: GELU</title><rect class="b c-act" x="326" y="933" width="528" height="22" rx="9"/><text class="box-label sm" x="590.0" y="948.5" text-anchor="middle">gelu GELU</text></g> |
| <g class=""><title>layernorm_final: LayerNorm 768</title><rect class="b c-norm" x="310" y="988" width="276" height="22" rx="9"/><text class="box-label sm" x="448.0" y="1003.5" text-anchor="middle">layernorm_final LayerNorm 768</text></g> |
| <g class=""><title>lm_head: Linear [768→10048]</title><rect class="b c-head" x="593" y="988" width="276" height="22" rx="9"/><text class="box-label sm" x="731.0" y="1003.5" text-anchor="middle">lm_head [768→10048]</text></g> |
| <g class=""><title>input_embeds_layers: 8× Embedding</title><rect class="b c-embed" x="310" y="1075" width="560" height="22" rx="9"/><text class="box-label sm" x="590.0" y="1090.5" text-anchor="middle">input_embeds_layers 8× Embedding</text></g> |
| <g class=""><title>position_embeds_layer: Embedding [1024×768]</title><rect class="b c-embed" x="310" y="1104" width="560" height="22" rx="9"/><text class="box-label sm" x="590.0" y="1119.5" text-anchor="middle">position_embeds_layer Embedding [1024×768]</text></g> |
| <g class=""><title>drop: Dropout</title><rect class="b c-sub" x="310" y="1133" width="560" height="22" rx="9"/><text class="box-label sm" x="590.0" y="1148.5" text-anchor="middle">drop Dropout</text></g> |
| <g class=""><title>layernorm_1: LayerNorm 768</title><rect class="b c-norm" x="318" y="1187" width="268" height="22" rx="9"/><text class="box-label sm" x="452.0" y="1202.5" text-anchor="middle">layernorm_1 LayerNorm 768</text></g> |
| <g class=""><title>layernorm_2: LayerNorm 768</title><rect class="b c-norm" x="593" y="1187" width="268" height="22" rx="9"/><text class="box-label sm" x="727.0" y="1202.5" text-anchor="middle">layernorm_2 LayerNorm 768</text></g> |
| <g class=""><title>attn_dropout: Dropout</title><rect class="b c-sub" x="326" y="1241" width="171" height="22" rx="9"/><text class="box-label sm" x="411.5" y="1256.5" text-anchor="middle">attn_dropout Dropout</text></g> |
| <g class=""><title>resid_dropout: Dropout</title><rect class="b c-sub" x="504" y="1241" width="171" height="22" rx="9"/><text class="box-label sm" x="589.5" y="1256.5" text-anchor="middle">resid_dropout Dropout</text></g> |
| <g class=""><title>att_proj: Linear [768→2304]</title><rect class="b c-proj" x="682" y="1241" width="171" height="22" rx="9"/><text class="box-label sm" x="767.5" y="1256.5" text-anchor="middle">att_proj [768→2304]</text></g> |
| <g class=""><title>out_proj: Linear [768→768]</title><rect class="b c-proj" x="326" y="1270" width="528" height="22" rx="9"/><text class="box-label sm" x="590.0" y="1285.5" text-anchor="middle">out_proj [768→768]</text></g> |
| <g class=""><title>in_proj: Linear [768→3072]</title><rect class="b c-proj" x="326" y="1337" width="171" height="22" rx="9"/><text class="box-label sm" x="411.5" y="1352.5" text-anchor="middle">in_proj [768→3072]</text></g> |
| <g class=""><title>out_proj: Linear [3072→768]</title><rect class="b c-proj" x="504" y="1337" width="171" height="22" rx="9"/><text class="box-label sm" x="589.5" y="1352.5" text-anchor="middle">out_proj [3072→768]</text></g> |
| <g class=""><title>dropout: Dropout</title><rect class="b c-sub" x="682" y="1337" width="171" height="22" rx="9"/><text class="box-label sm" x="767.5" y="1352.5" text-anchor="middle">dropout Dropout</text></g> |
| <g class=""><title>gelu: GELU</title><rect class="b c-act" x="326" y="1366" width="528" height="22" rx="9"/><text class="box-label sm" x="590.0" y="1381.5" text-anchor="middle">gelu GELU</text></g> |
| <g class=""><title>layernorm_final: LayerNorm 768</title><rect class="b c-norm" x="310" y="1421" width="276" height="22" rx="9"/><text class="box-label sm" x="448.0" y="1436.5" text-anchor="middle">layernorm_final LayerNorm 768</text></g> |
| <g class=""><title>lm_heads: 7× Linear</title><rect class="b c-proj" x="593" y="1421" width="276" height="22" rx="9"/><text class="box-label sm" x="731.0" y="1436.5" text-anchor="middle">lm_heads 7× Linear</text></g> |
| <g class=""><title>conv: ParametrizedConv1d 1→32 k7</title><rect class="b c-conv" x="342" y="1608" width="496" height="22" rx="9"/><text class="box-label sm" x="590.0" y="1623.5" text-anchor="middle">conv ParametrizedConv1d 1→32 k7</text></g> |
| <g class=""><title>0: ELU</title><rect class="b c-sub" x="358" y="1725" width="464" height="22" rx="9"/><text class="box-label sm" x="590.0" y="1740.5" text-anchor="middle">0 ELU</text></g> |
| <g class=""><title>conv: ParametrizedConv1d 32→16 k3</title><rect class="b c-conv" x="366" y="1779" width="448" height="22" rx="9"/><text class="box-label sm" x="590.0" y="1794.5" text-anchor="middle">conv ParametrizedConv1d 32→16 k3</text></g> |
| <g class=""><title>conv: ParametrizedConv1d 32→32 k1</title><rect class="b c-conv" x="350" y="1872" width="480" height="22" rx="9"/><text class="box-label sm" x="590.0" y="1887.5" text-anchor="middle">conv ParametrizedConv1d 32→32 k1</text></g> |
| <g class=""><title>2: ELU</title><rect class="b c-sub" x="334" y="1927" width="512" height="22" rx="9"/><text class="box-label sm" x="590.0" y="1942.5" text-anchor="middle">2 ELU</text></g> |
| <g class=""><title>conv: ParametrizedConv1d 256→512 k16</title><rect class="b c-conv" x="334" y="1994" width="512" height="22" rx="9"/><text class="box-label sm" x="590.0" y="2009.5" text-anchor="middle">conv ParametrizedConv1d 256→512 k16</text></g> |
| <g class=""><title>lstm: LSTM</title><rect class="b c-recur" x="334" y="2061" width="512" height="22" rx="9"/><text class="box-label sm" x="590.0" y="2076.5" text-anchor="middle">lstm LSTM</text></g> |
| <g class=""><title>14: ELU</title><rect class="b c-sub" x="326" y="2103" width="528" height="22" rx="9"/><text class="box-label sm" x="590.0" y="2118.5" text-anchor="middle">14 ELU</text></g> |
| <g class=""><title>conv: ParametrizedConv1d 512→128 k7</title><rect class="b c-conv" x="334" y="2157" width="512" height="22" rx="9"/><text class="box-label sm" x="590.0" y="2172.5" text-anchor="middle">conv ParametrizedConv1d 512→128 k7</text></g> |
| <g class=""><title>conv: ParametrizedConv1d 128→512 k7</title><rect class="b c-conv" x="334" y="2300" width="512" height="22" rx="9"/><text class="box-label sm" x="590.0" y="2315.5" text-anchor="middle">conv ParametrizedConv1d 128→512 k7</text></g> |
| <g class=""><title>lstm: LSTM</title><rect class="b c-recur" x="334" y="2367" width="512" height="22" rx="9"/><text class="box-label sm" x="590.0" y="2382.5" text-anchor="middle">lstm LSTM</text></g> |
| <g class=""><title>2: ELU</title><rect class="b c-sub" x="334" y="2434" width="512" height="22" rx="9"/><text class="box-label sm" x="590.0" y="2449.5" text-anchor="middle">2 ELU</text></g> |
| <g class=""><title>parametrizations: ModuleDict</title><rect class="b c-sub" x="350" y="2513" width="480" height="22" rx="9"/><text class="box-label sm" x="590.0" y="2528.5" text-anchor="middle">parametrizations ModuleDict</text></g> |
| <g class=""><title>0: ELU</title><rect class="b c-sub" x="358" y="2643" width="464" height="22" rx="9"/><text class="box-label sm" x="590.0" y="2658.5" text-anchor="middle">0 ELU</text></g> |
| <g class=""><title>conv: ParametrizedConv1d 256→128 k3</title><rect class="b c-conv" x="366" y="2697" width="448" height="22" rx="9"/><text class="box-label sm" x="590.0" y="2712.5" text-anchor="middle">conv ParametrizedConv1d 256→128 k3</text></g> |
| <g class=""><title>conv: ParametrizedConv1d 256→256 k1</title><rect class="b c-conv" x="350" y="2790" width="480" height="22" rx="9"/><text class="box-label sm" x="590.0" y="2805.5" text-anchor="middle">conv ParametrizedConv1d 256→256 k1</text></g> |
| <g class=""><title>14: ELU</title><rect class="b c-sub" x="326" y="2858" width="528" height="22" rx="9"/><text class="box-label sm" x="590.0" y="2873.5" text-anchor="middle">14 ELU</text></g> |
| <g class=""><title>conv: ParametrizedConv1d 32→1 k7</title><rect class="b c-conv" x="334" y="2912" width="512" height="22" rx="9"/><text class="box-label sm" x="590.0" y="2927.5" text-anchor="middle">conv ParametrizedConv1d 32→1 k7</text></g> |
| <g class=""><title>codebook: EncodecEuclideanCodebook</title><rect class="b c-quant" x="326" y="3030" width="528" height="22" rx="9"/><text class="box-label sm" x="590.0" y="3045.5" text-anchor="middle">codebook EncodecEuclideanCodebook</text></g> |
| <rect class="panel" x="912" y="88" width="264" height="328" rx="8"/> |
| <text class="facts-k" x="926" y="112">model id</text> |
| <text class="facts-v" x="1162" y="112" text-anchor="end">suno/bark</text> |
| <text class="facts-k" x="926" y="134">model_type</text> |
| <text class="facts-v" x="1162" y="134" text-anchor="end">bark</text> |
| <text class="facts-k" x="926" y="156">config</text> |
| <text class="facts-v" x="1162" y="156" text-anchor="end">BarkConfig</text> |
| <text class="facts-k" x="926" y="178">decoder</text> |
| <text class="facts-v" x="1162" y="178" text-anchor="end">dense</text> |
| <text class="facts-k" x="926" y="200">layers</text> |
| <text class="facts-v" x="1162" y="200" text-anchor="end">32</text> |
| <text class="facts-k" x="926" y="222">hidden</text> |
| <text class="facts-v" x="1162" y="222" text-anchor="end">—</text> |
| <text class="facts-k" x="926" y="244">intermediate</text> |
| <text class="facts-v" x="1162" y="244" text-anchor="end">—</text> |
| <text class="facts-k" x="926" y="266">heads / kv</text> |
| <text class="facts-v" x="1162" y="266" text-anchor="end">None / None</text> |
| <text class="facts-k" x="926" y="288">attention</text> |
| <text class="facts-v" x="1162" y="288" text-anchor="end">—</text> |
| <text class="facts-k" x="926" y="310">norm</text> |
| <text class="facts-v" x="1162" y="310" text-anchor="end">LayerNorm</text> |
| <text class="facts-k" x="926" y="332">positional</text> |
| <text class="facts-v" x="1162" y="332" text-anchor="end">n/a</text> |
| <text class="facts-k" x="926" y="354">vocab</text> |
| <text class="facts-v" x="1162" y="354" text-anchor="end">—</text> |
| <text class="facts-k" x="926" y="376">max_pos</text> |
| <text class="facts-v" x="1162" y="376" text-anchor="end">—</text> |
| <text class="facts-k" x="926" y="398">tie_embeddings</text> |
| <text class="facts-v" x="1162" y="398" text-anchor="end">None</text> |
| <text class="legend-t" x="914" y="436" font-weight="700">legend</text> |
| <rect x="916" y="447" width="16" height="12" rx="2" fill="var(--embed-s)"/> |
| <text class="legend-t" x="940" y="457">embedding</text> |
| <rect x="916" y="465" width="16" height="12" rx="2" fill="var(--attn-s)"/> |
| <text class="legend-t" x="940" y="475">attention</text> |
| <rect x="916" y="483" width="16" height="12" rx="2" fill="var(--mamba-s)"/> |
| <text class="legend-t" x="940" y="493">Mamba / SSM</text> |
| <rect x="916" y="501" width="16" height="12" rx="2" fill="var(--moe-s)"/> |
| <text class="legend-t" x="940" y="511">MoE / experts</text> |
| <rect x="916" y="519" width="16" height="12" rx="2" fill="var(--proj-s)"/> |
| <text class="legend-t" x="940" y="529">linear / proj</text> |
| <rect x="916" y="537" width="16" height="12" rx="2" fill="var(--grid)"/> |
| <text class="legend-t" x="940" y="547">convolution</text> |
| <rect x="916" y="555" width="16" height="12" rx="2" fill="var(--norm-s)"/> |
| <text class="legend-t" x="940" y="565">normalization</text> |
| <rect x="916" y="573" width="16" height="12" rx="2" fill="var(--grid)"/> |
| <text class="legend-t" x="940" y="583">activation</text> |
| <rect x="916" y="591" width="16" height="12" rx="2" fill="var(--grid)"/> |
| <text class="legend-t" x="940" y="601">pooling</text> |
| <rect x="916" y="609" width="16" height="12" rx="2" fill="var(--head-s)"/> |
| <text class="legend-t" x="940" y="619">head</text> |
| </svg> |
