diff --git a/docs/src/snippets/configs/meta-models/muse-glimmer.jsx b/docs/src/snippets/configs/meta-models/muse-glimmer.jsx index 27e77548b..563a1abf9 100644 --- a/docs/src/snippets/configs/meta-models/muse-glimmer.jsx +++ b/docs/src/snippets/configs/meta-models/muse-glimmer.jsx @@ -127,8 +127,7 @@ export const config = { { id: "off", label: "Off" }, { id: "dflash", label: "DFlash", flags: ["--speculative-algorithm DFLASH", - "--speculative-draft-model-path {{DRAFT_PATH}}", - "--speculative-dflash-block-size 5"] }, + "--speculative-draft-model-path {{DRAFT_PATH}}"] }, ], }, }, @@ -164,7 +163,6 @@ export const config = { "--speculative-algorithm DFLASH", "--speculative-draft-model-path {{DRAFT_PATH}}", "--speculative-draft-load-format auto", - "--speculative-dflash-block-size 5", "--mem-fraction-static 0.85", "--host {{HOST_IP}}", "--port {{PORT}}", @@ -214,7 +212,6 @@ export const config = { "--tool-call-parser muse", "--speculative-algorithm DFLASH", "--speculative-draft-model-path {{DRAFT_PATH}}", - "--speculative-dflash-block-size 5", "--speculative-draft-model-quantization fp8", "--kv-cache-dtype fp8_e4m3", "--mem-fraction-static 0.9", @@ -246,7 +243,6 @@ export const config = { "--tool-call-parser muse", "--speculative-algorithm DFLASH", "--speculative-draft-model-path {{DRAFT_PATH}}", - "--speculative-dflash-block-size 5", "--mem-fraction-static 0.85", "--host {{HOST_IP}}", "--port {{PORT}}", @@ -286,7 +282,6 @@ export const config = { "--tool-call-parser muse", "--speculative-algorithm DFLASH", "--speculative-draft-model-path {{DRAFT_PATH}}", - "--speculative-dflash-block-size 5", "--speculative-draft-model-quantization fp8", "--kv-cache-dtype fp8_e4m3", "--mem-fraction-static 0.85", @@ -322,7 +317,6 @@ export const config = { "--tool-call-parser muse", "--speculative-algorithm DFLASH", "--speculative-draft-model-path {{DRAFT_PATH}}", - "--speculative-dflash-block-size 5", "--mem-fraction-static 0.65", "--host {{HOST_IP}}", "--port {{PORT}}", @@ -362,7 +356,6 @@ export const config = { "--tool-call-parser muse", "--speculative-algorithm DFLASH", "--speculative-draft-model-path {{DRAFT_PATH}}", - "--speculative-dflash-block-size 5", "--speculative-draft-model-quantization fp8", "--kv-cache-dtype fp8_e4m3", "--mem-fraction-static 0.38", @@ -394,7 +387,6 @@ export const config = { "--tool-call-parser muse", "--speculative-algorithm DFLASH", "--speculative-draft-model-path {{DRAFT_PATH}}", - "--speculative-dflash-block-size 5", "--mem-fraction-static 0.85", "--host {{HOST_IP}}", "--port {{PORT}}", @@ -470,7 +462,6 @@ export const config = { "--tool-call-parser muse", "--speculative-algorithm DFLASH", "--speculative-draft-model-path {{DRAFT_PATH}}", - "--speculative-dflash-block-size 5", "--mem-fraction-static 0.85", "--host {{HOST_IP}}", "--port {{PORT}}", @@ -510,7 +501,6 @@ export const config = { "--tool-call-parser muse", "--speculative-algorithm DFLASH", "--speculative-draft-model-path {{DRAFT_PATH}}", - "--speculative-dflash-block-size 5", "--speculative-draft-model-quantization fp8", "--kv-cache-dtype fp8_e4m3", "--mem-fraction-static 0.85",