diff --git a/modules/services/ollama.nix b/modules/services/ollama.nix index 8e4c52178..1fb430898 100644 --- a/modules/services/ollama.nix +++ b/modules/services/ollama.nix @@ -59,6 +59,7 @@ let llama3_2-uncensored-3b # llama3_3-70b # non-compliant; dodges iffy questions llama3_3-abliterated-70b # compliant, but slower and not as helpful as deepseek-r1-abliterated-70b + llama4-16x17b magicoder-7b # it generates valid, if sparse, code magistral-24b marco-o1-7b # untested diff --git a/pkgs/ollamaPackages/README.md b/pkgs/ollamaPackages/README.md index b082e8cae..c69d9e6da 100644 --- a/pkgs/ollamaPackages/README.md +++ b/pkgs/ollamaPackages/README.md @@ -78,7 +78,7 @@ ollama API isn't documented anywhere, and it has changed over time, but it's all - released 2025-04-07 - - -- [ ] llama4 (16x17b, 128x17b) +- [x] llama4 (16x17b, 128x17b) - released 2025-04-05 - - diff --git a/pkgs/ollamaPackages/llama4-16x17b/package.nix b/pkgs/ollamaPackages/llama4-16x17b/package.nix new file mode 100644 index 000000000..3cff440b3 --- /dev/null +++ b/pkgs/ollamaPackages/llama4-16x17b/package.nix @@ -0,0 +1,14 @@ +# +# +# released 2025-04-05 +{ mkOllamaModel }: mkOllamaModel { + modelName = "llama4"; + variant = "16x17b"; + manifestHash = "sha256-vzFgTiXCXZZOJQvPKKgr+9voivXyNiV/q7J2Kbskx/M="; + modelBlob = "9d507a36062c2845dd3bb3e93364e9abc1607118acd8650727a700f72fb126e5"; + modelBlobHash = "sha256-nVB6NgYsKEXdO7PpM2Tpq8FgcRis2GUHJ6cA9y+xJuU="; + paramsBlob = "bee89e20d457c132784e74ae48177b45262ecc7383c085c835ec54da40d2e4e6"; + paramsBlobHash = "sha256-vuieINRXwTJ4TnSuSBd7RSYuzHODwIXINexU2kDS5OY="; + systemBlob = "fc1ffc71ab8ebabe8ec0177ea8ee41d1ea27db856636a517d54324eecdfb3f11"; + systemBlobHash = "sha256-/B/8cauOur6OwBd+qO5B0eon24VmNqUX1UMk7s37PxE="; +}