update llama.cpp submodule to c29af7e (#2868 )

Fix embeddings load model behavior (#2848 )
Merge pull request #2838 from dhiltgen/opensuse
2024-03-01 15:26:04 -08:00 · 2024-02-29 17:40:56 -08:00 · 2024-02-29 15:47:56 -08:00 · 2024-02-29 15:47:37 -08:00 · 2024-02-29 15:46:46 -08:00 · 2024-02-29 12:12:13 -08:00
26 changed files with 121 additions and 241 deletions
--- a/1
+++ b/1
@ -132,6 +132,7 @@ ENV OLLAMA_HOST 0.0.0.0
 ENV PATH=/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin
 ENV LD_LIBRARY_PATH=/usr/local/nvidia/lib:/usr/local/nvidia/lib64
 ENV NVIDIA_DRIVER_CAPABILITIES=compute,utility
+ENV NVIDIA_VISIBLE_DEVICES=all

 ENTRYPOINT ["/bin/ollama"]
 CMD ["serve"]
--- a/README.md
+++ b/README.md
@ -275,6 +275,8 @@ See the [API documentation](./docs/api.md) for all endpoints.
 - [MindMac](https://mindmac.app)
 - [NextJS Web Interface for Ollama](https://github.com/jakobhoeg/nextjs-ollama-llm-ui)
 - [Msty](https://msty.app)
+- [Chatbox](https://github.com/Bin-Huang/Chatbox)
+- [NextChat](https://github.com/ChatGPTNextWeb/ChatGPT-Next-Web) with [Get Started Doc](https://docs.nextchat.dev/models/ollama)

 ### Terminal

--- a/api/types.go
+++ b/api/types.go
@ -83,7 +83,7 @@ type Metrics struct {
 	EvalDuration       time.Duration `json:"eval_duration,omitempty"`
 }

-// Options specfied in GenerateRequest, if you add a new option here add it to the API docs also
+// Options specified in GenerateRequest, if you add a new option here add it to the API docs also
 type Options struct {
 	Runner

@ -121,7 +121,6 @@ type Runner struct {
 	VocabOnly          bool    `json:"vocab_only,omitempty"`
 	UseMMap            bool    `json:"use_mmap,omitempty"`
 	UseMLock           bool    `json:"use_mlock,omitempty"`
-	EmbeddingOnly      bool    `json:"embedding_only,omitempty"`
 	RopeFrequencyBase  float32 `json:"rope_frequency_base,omitempty"`
 	RopeFrequencyScale float32 `json:"rope_frequency_scale,omitempty"`
 	NumThread          int     `json:"num_thread,omitempty"`
@ -395,7 +394,6 @@ func DefaultOptions() Options {
 			UseMLock:           false,
 			UseMMap:            true,
 			UseNUMA:            false,
-			EmbeddingOnly:      true,
 		},
 	}
 }
--- a/app/lifecycle/updater.go
+++ b/app/lifecycle/updater.go
@ -86,6 +86,11 @@ func IsNewReleaseAvailable(ctx context.Context) (bool, UpdateResponse) {
 	if err != nil {
 		slog.Warn(fmt.Sprintf("failed to read body response: %s", err))
 	}
+
+	if resp.StatusCode != 200 {
+		slog.Info(fmt.Sprintf("check update error %d - %.96s", resp.StatusCode, string(body)))
+		return false, updateResp
+	}
 	err = json.Unmarshal(body, &updateResp)
 	if err != nil {
 		slog.Warn(fmt.Sprintf("malformed response checking for update: %s", err))
--- a/app/ollama.iss
+++ b/app/ollama.iss
@ -37,7 +37,7 @@ PrivilegesRequired=lowest
 OutputBaseFilename="OllamaSetup"
 SetupIconFile={#MyIcon}
 UninstallDisplayIcon={uninstallexe}
-Compression=zip
+Compression=lzma2
 SolidCompression=no
 WizardStyle=modern
 ChangesEnvironment=yes
@ -49,9 +49,6 @@ SetupLogging=yes
 CloseApplications=yes
 RestartApplications=no

-; Make sure they can at least download llama2 as a minimum
-ExtraDiskSpaceRequired=3826806784
-
 ; https://jrsoftware.org/ishelp/index.php?topic=setup_wizardimagefile
 WizardSmallImageFile=.\assets\setup.bmp

@ -116,7 +113,8 @@ Filename: "{cmd}"; Parameters: "/c timeout 5"; Flags: runhidden
 Type: filesandordirs; Name: "{%TEMP}\ollama*"
 Type: filesandordirs; Name: "{%LOCALAPPDATA}\Ollama"
 Type: filesandordirs; Name: "{%LOCALAPPDATA}\Programs\Ollama"
-Type: filesandordirs; Name: "{%USERPROFILE}\.ollama"
+Type: filesandordirs; Name: "{%USERPROFILE}\.ollama\models"
+Type: filesandordirs; Name: "{%USERPROFILE}\.ollama\history"
 ; NOTE: if the user has a custom OLLAMA_MODELS it will be preserved

 [Messages]
--- a/cmd/cmd.go
+++ b/cmd/cmd.go
@ -718,39 +718,36 @@ func initializeKeypair() error {
 	_, err = os.Stat(privKeyPath)
 	if os.IsNotExist(err) {
 		fmt.Printf("Couldn't find '%s'. Generating new private key.\n", privKeyPath)
-		_, privKey, err := ed25519.GenerateKey(rand.Reader)
+		cryptoPublicKey, cryptoPrivateKey, err := ed25519.GenerateKey(rand.Reader)
 		if err != nil {
 			return err
 		}

-		privKeyBytes, err := format.OpenSSHPrivateKey(privKey, "")
+		privateKeyBytes, err := ssh.MarshalPrivateKey(cryptoPrivateKey, "")
 		if err != nil {
 			return err
 		}

-		err = os.MkdirAll(filepath.Dir(privKeyPath), 0o755)
-		if err != nil {
+		if err := os.MkdirAll(filepath.Dir(privKeyPath), 0o755); err != nil {
 			return fmt.Errorf("could not create directory %w", err)
 		}

-		err = os.WriteFile(privKeyPath, pem.EncodeToMemory(privKeyBytes), 0o600)
+		if err := os.WriteFile(privKeyPath, pem.EncodeToMemory(privateKeyBytes), 0o600); err != nil {
+			return err
+		}
+
+		sshPublicKey, err := ssh.NewPublicKey(cryptoPublicKey)
 		if err != nil {
 			return err
 		}

-		sshPrivateKey, err := ssh.NewSignerFromKey(privKey)
-		if err != nil {
+		publicKeyBytes := ssh.MarshalAuthorizedKey(sshPublicKey)
+
+		if err := os.WriteFile(pubKeyPath, publicKeyBytes, 0o644); err != nil {
 			return err
 		}

-		pubKeyData := ssh.MarshalAuthorizedKey(sshPrivateKey.PublicKey())
-
-		err = os.WriteFile(pubKeyPath, pubKeyData, 0o644)
-		if err != nil {
-			return err
-		}
-
-		fmt.Printf("Your new public key is: \n\n%s\n", string(pubKeyData))
+		fmt.Printf("Your new public key is: \n\n%s\n", publicKeyBytes)
 	}
 	return nil
 }
--- a/docs/api.md
+++ b/docs/api.md
@ -321,7 +321,6 @@ curl http://localhost:11434/api/generate -d '{
    "vocab_only": false,
    "use_mmap": true,
    "use_mlock": false,
-    "embedding_only": false,
    "rope_frequency_base": 1.1,
    "rope_frequency_scale": 0.8,
    "num_thread": 8
--- a/docs/tutorials/langchainpy.md
+++ b/docs/tutorials/langchainpy.md
@ -42,12 +42,12 @@ text_splitter=RecursiveCharacterTextSplitter(chunk_size=500, chunk_overlap=0)
 all_splits = text_splitter.split_documents(data)
 ```

-It's split up, but we have to find the relevant splits and then submit those to the model. We can do this by creating embeddings and storing them in a vector database. We can use Ollama directly to instantiate an embedding model. We will use ChromaDB in this example for a vector database. `pip install GPT4All chromadb`
+It's split up, but we have to find the relevant splits and then submit those to the model. We can do this by creating embeddings and storing them in a vector database. We can use Ollama directly to instantiate an embedding model. We will use ChromaDB in this example for a vector database. `pip install chromadb`

 ```python
 from langchain.embeddings import OllamaEmbeddings
 from langchain.vectorstores import Chroma
-oembed = OllamaEmbeddings(base_url="http://localhost:11434", model="llama2")
+oembed = OllamaEmbeddings(base_url="http://localhost:11434", model="nomic-embed-text")
 vectorstore = Chroma.from_documents(documents=all_splits, embedding=oembed)
 ```

@ -66,7 +66,7 @@ The next thing is to send the question and the relevant parts of the docs to the
 ```python
 from langchain.chains import RetrievalQA
 qachain=RetrievalQA.from_chain_type(ollama, retriever=vectorstore.as_retriever())
-qachain({"query": question})
+qachain.invoke({"query": question})
 ```

 The answer received from this chain was:
--- a/format/openssh.go
+++ b/format/openssh.go
@ -1,102 +0,0 @@
-// Copyright 2012 The Go Authors. All rights reserved.
-// Use of this source code is governed by a BSD-style
-// license that can be found in the LICENSE file.
-
-// Code originally from https://go-review.googlesource.com/c/crypto/+/218620
-
-// TODO: replace with upstream once the above change is merged and released.
-
-package format
-
-import (
-	"crypto"
-	"crypto/ed25519"
-	"crypto/rand"
-	"encoding/binary"
-	"encoding/pem"
-	"fmt"
-
-	"golang.org/x/crypto/ssh"
-)
-
-const privateKeyAuthMagic = "openssh-key-v1\x00"
-
-type openSSHEncryptedPrivateKey struct {
-	CipherName string
-	KDFName    string
-	KDFOptions string
-	KeysCount  uint32
-	PubKey     []byte
-	KeyBlocks  []byte
-}
-
-type openSSHPrivateKey struct {
-	Check1  uint32
-	Check2  uint32
-	Keytype string
-	Rest    []byte `ssh:"rest"`
-}
-
-type openSSHEd25519PrivateKey struct {
-	Pub     []byte
-	Priv    []byte
-	Comment string
-	Pad     []byte `ssh:"rest"`
-}
-
-func OpenSSHPrivateKey(key crypto.PrivateKey, comment string) (*pem.Block, error) {
-	var check uint32
-	if err := binary.Read(rand.Reader, binary.BigEndian, &check); err != nil {
-		return nil, err
-	}
-
-	var pk1 openSSHPrivateKey
-	pk1.Check1 = check
-	pk1.Check2 = check
-
-	var w openSSHEncryptedPrivateKey
-	w.KeysCount = 1
-
-	if k, ok := key.(*ed25519.PrivateKey); ok {
-		key = *k
-	}
-
-	switch k := key.(type) {
-	case ed25519.PrivateKey:
-		pub, priv := k[32:], k
-		key := openSSHEd25519PrivateKey{
-			Pub:     pub,
-			Priv:    priv,
-			Comment: comment,
-		}
-
-		pk1.Keytype = ssh.KeyAlgoED25519
-		pk1.Rest = ssh.Marshal(key)
-
-		w.PubKey = ssh.Marshal(struct {
-			KeyType string
-			Pub     []byte
-		}{
-			ssh.KeyAlgoED25519, pub,
-		})
-	default:
-		return nil, fmt.Errorf("ssh: unknown key type %T", k)
-	}
-
-	w.KeyBlocks = openSSHPadding(ssh.Marshal(pk1), 8)
-
-	w.CipherName, w.KDFName, w.KDFOptions = "none", "none", ""
-
-	return &pem.Block{
-		Type:  "OPENSSH PRIVATE KEY",
-		Bytes: append([]byte(privateKeyAuthMagic), ssh.Marshal(w)...),
-	}, nil
-}
-
-func openSSHPadding(block []byte, blocksize int) []byte {
-	for i, j := 0, len(block); (j+i)%blocksize != 0; i++ {
-		block = append(block, byte(i+1))
-	}
-
-	return block
-}
--- a/go.mod
+++ b/go.mod
@ -3,8 +3,10 @@ module github.com/jmorganca/ollama
 go 1.21

 require (
+	github.com/containerd/console v1.0.3
 	github.com/emirpasic/gods v1.18.1
 	github.com/gin-gonic/gin v1.9.1
+	github.com/google/uuid v1.0.0
 	github.com/olekukonko/tablewriter v0.0.5
 	github.com/spf13/cobra v1.7.0
 	github.com/stretchr/testify v1.8.4
@ -12,26 +14,12 @@ require (
 )

 require (
-	github.com/containerd/console v1.0.3 // indirect
-	github.com/cratonica/2goarray v0.0.0-20190331194516-514510793eaa // indirect
 	github.com/davecgh/go-spew v1.1.1 // indirect
-	github.com/getlantern/context v0.0.0-20190109183933-c447772a6520 // indirect
-	github.com/getlantern/errors v0.0.0-20190325191628-abdb3e3e36f7 // indirect
-	github.com/getlantern/golog v0.0.0-20190830074920-4ef2e798c2d7 // indirect
-	github.com/getlantern/hex v0.0.0-20190417191902-c6586a6fe0b7 // indirect
-	github.com/getlantern/hidden v0.0.0-20190325191715-f02dbb02be55 // indirect
-	github.com/getlantern/ops v0.0.0-20190325191751-d70cb0d6f85f // indirect
-	github.com/getlantern/systray v1.2.2 // indirect
-	github.com/go-stack/stack v1.8.0 // indirect
-	github.com/google/uuid v1.0.0 // indirect
 	github.com/mattn/go-runewidth v0.0.14 // indirect
-	github.com/oxtoacart/bpool v0.0.0-20190530202638-03653db5a59c // indirect
-	github.com/pborman/uuid v1.2.1 // indirect
 	github.com/pmezard/go-difflib v1.0.0 // indirect
 	github.com/rivo/uniseg v0.2.0 // indirect
 )

-
 require (
 	github.com/bytedance/sonic v1.9.1 // indirect
 	github.com/chenzhuoyu/base64x v0.0.0-20221115062448-fe3a3abad311 // indirect
--- a/go.sum
+++ b/go.sum
@ -7,8 +7,6 @@ github.com/chenzhuoyu/base64x v0.0.0-20221115062448-fe3a3abad311/go.mod h1:b583j
 github.com/containerd/console v1.0.3 h1:lIr7SlA5PxZyMV30bDW0MGbiOPXwc63yRuCP0ARubLw=
 github.com/containerd/console v1.0.3/go.mod h1:7LqA/THxQ86k76b8c/EMSiaJ3h1eZkMkXar0TQ1gf3U=
 github.com/cpuguy83/go-md2man/v2 v2.0.2/go.mod h1:tgQtvFlXSQOSOSIRvRPT7W67SCa46tRHOmNcaadrF8o=
-github.com/cratonica/2goarray v0.0.0-20190331194516-514510793eaa h1:Wg+722vs7a2zQH5lR9QWYsVbplKeffaQFIs5FTdfNNo=
-github.com/cratonica/2goarray v0.0.0-20190331194516-514510793eaa/go.mod h1:6Arca19mRx58CA7OWEd7Wu1NpC1rd3uDnNs6s1pj/DI=
 github.com/creack/pty v1.1.9/go.mod h1:oKZEueFk5CKHvIhNR5MUki03XCEU+Q6VDXinZuGJ33E=
 github.com/davecgh/go-spew v1.1.0/go.mod h1:J7Y8YcW2NihsgmVo/mv3lAwl/skON4iLHjSsI+c5H38=
 github.com/davecgh/go-spew v1.1.1 h1:vj9j/u1bqnvCEfJOwUhtlOARqs3+rkHYY13jYWTU97c=
@ -17,20 +15,6 @@ github.com/emirpasic/gods v1.18.1 h1:FXtiHYKDGKCW2KzwZKx0iC0PQmdlorYgdFG9jPXJ1Bc
 github.com/emirpasic/gods v1.18.1/go.mod h1:8tpGGwCnJ5H4r6BWwaV6OrWmMoPhUl5jm/FMNAnJvWQ=
 github.com/gabriel-vasile/mimetype v1.4.2 h1:w5qFW6JKBz9Y393Y4q372O9A7cUSequkh1Q7OhCmWKU=
 github.com/gabriel-vasile/mimetype v1.4.2/go.mod h1:zApsH/mKG4w07erKIaJPFiX0Tsq9BFQgN3qGY5GnNgA=
-github.com/getlantern/context v0.0.0-20190109183933-c447772a6520 h1:NRUJuo3v3WGC/g5YiyF790gut6oQr5f3FBI88Wv0dx4=
-github.com/getlantern/context v0.0.0-20190109183933-c447772a6520/go.mod h1:L+mq6/vvYHKjCX2oez0CgEAJmbq1fbb/oNJIWQkBybY=
-github.com/getlantern/errors v0.0.0-20190325191628-abdb3e3e36f7 h1:6uJ+sZ/e03gkbqZ0kUG6mfKoqDb4XMAzMIwlajq19So=
-github.com/getlantern/errors v0.0.0-20190325191628-abdb3e3e36f7/go.mod h1:l+xpFBrCtDLpK9qNjxs+cHU6+BAdlBaxHqikB6Lku3A=
-github.com/getlantern/golog v0.0.0-20190830074920-4ef2e798c2d7 h1:guBYzEaLz0Vfc/jv0czrr2z7qyzTOGC9hiQ0VC+hKjk=
-github.com/getlantern/golog v0.0.0-20190830074920-4ef2e798c2d7/go.mod h1:zx/1xUUeYPy3Pcmet8OSXLbF47l+3y6hIPpyLWoR9oc=
-github.com/getlantern/hex v0.0.0-20190417191902-c6586a6fe0b7 h1:micT5vkcr9tOVk1FiH8SWKID8ultN44Z+yzd2y/Vyb0=
-github.com/getlantern/hex v0.0.0-20190417191902-c6586a6fe0b7/go.mod h1:dD3CgOrwlzca8ed61CsZouQS5h5jIzkK9ZWrTcf0s+o=
-github.com/getlantern/hidden v0.0.0-20190325191715-f02dbb02be55 h1:XYzSdCbkzOC0FDNrgJqGRo8PCMFOBFL9py72DRs7bmc=
-github.com/getlantern/hidden v0.0.0-20190325191715-f02dbb02be55/go.mod h1:6mmzY2kW1TOOrVy+r41Za2MxXM+hhqTtY3oBKd2AgFA=
-github.com/getlantern/ops v0.0.0-20190325191751-d70cb0d6f85f h1:wrYrQttPS8FHIRSlsrcuKazukx/xqO/PpLZzZXsF+EA=
-github.com/getlantern/ops v0.0.0-20190325191751-d70cb0d6f85f/go.mod h1:D5ao98qkA6pxftxoqzibIBBrLSUli+kYnJqrgBf9cIA=
-github.com/getlantern/systray v1.2.2 h1:dCEHtfmvkJG7HZ8lS/sLklTH4RKUcIsKrAD9sThoEBE=
-github.com/getlantern/systray v1.2.2/go.mod h1:pXFOI1wwqwYXEhLPm9ZGjS2u/vVELeIgNMY5HvhHhcE=
 github.com/gin-contrib/cors v1.4.0 h1:oJ6gwtUl3lqV0WEIwM/LxPF1QZ5qe2lGWdY2+bz7y0g=
 github.com/gin-contrib/cors v1.4.0/go.mod h1:bs9pNM0x/UsmHPBWT2xZz9ROh8xYjYkiURUfmBoMlcs=
 github.com/gin-contrib/sse v0.1.0 h1:Y/yl/+YNO8GZSjAhjMsSuLt29uWRFHdHYUb5lYOV9qE=
@ -40,6 +24,7 @@ github.com/gin-gonic/gin v1.9.1 h1:4idEAncQnU5cB7BeOkPtxjfCSye0AAm1R0RVIqJ+Jmg=
 github.com/gin-gonic/gin v1.9.1/go.mod h1:hPrL7YrpYKXt5YId3A/Tnip5kqbEAP+KLuI3SUcPTeU=
 github.com/go-playground/assert/v2 v2.0.1/go.mod h1:VDjEfimB/XKnb+ZQfWdccd7VUvScMdVu0Titje2rxJ4=
 github.com/go-playground/assert/v2 v2.2.0 h1:JvknZsQTYeFEAhQwI4qEt9cyV5ONwRHC+lYKSsYSR8s=
+github.com/go-playground/assert/v2 v2.2.0/go.mod h1:VDjEfimB/XKnb+ZQfWdccd7VUvScMdVu0Titje2rxJ4=
 github.com/go-playground/locales v0.14.0/go.mod h1:sawfccIbzZTqEDETgFXqTho0QybSa7l++s0DH+LDiLs=
 github.com/go-playground/locales v0.14.1 h1:EWaQ/wswjilfKLTECiXz7Rh+3BjFhfDFKv/oXslEjJA=
 github.com/go-playground/locales v0.14.1/go.mod h1:hxrqLVvrK65+Rwrd5Fc6F2O76J/NuW9t0sjnWqG1slY=
@ -49,8 +34,6 @@ github.com/go-playground/universal-translator v0.18.1/go.mod h1:xekY+UJKNuX9WP91
 github.com/go-playground/validator/v10 v10.10.0/go.mod h1:74x4gJWsvQexRdW8Pn3dXSGrTK4nAUsbPlLADvpJkos=
 github.com/go-playground/validator/v10 v10.14.0 h1:vgvQWe3XCz3gIeFDm/HnTIbj6UGmg/+t63MyGU2n5js=
 github.com/go-playground/validator/v10 v10.14.0/go.mod h1:9iXMNT7sEkjXb0I+enO7QXmzG6QCsPWY4zveKFVRSyU=
-github.com/go-stack/stack v1.8.0 h1:5SgMzNM5HxrEjV0ww2lTmX6E2Izsfxas4+YHWRs3Lsk=
-github.com/go-stack/stack v1.8.0/go.mod h1:v0f6uXyyMGvRgIKkXu+yp6POWl0qKG85gN/melR3HDY=
 github.com/goccy/go-json v0.9.7/go.mod h1:6MelG93GURQebXPDq3khkgXZkazVtN9CRI+MGFi0w8I=
 github.com/goccy/go-json v0.10.2 h1:CrxCmQqYDkv1z7lO7Wbh2HN93uovUHgrECaO5ZrCXAU=
 github.com/goccy/go-json v0.10.2/go.mod h1:6MelG93GURQebXPDq3khkgXZkazVtN9CRI+MGFi0w8I=
@ -79,8 +62,6 @@ github.com/kr/text v0.2.0/go.mod h1:eLer722TekiGuMkidMxC/pM04lWEeraHUUmBw8l2grE=
 github.com/leodido/go-urn v1.2.1/go.mod h1:zt4jvISO2HfUBqxjfIshjdMTYS56ZS/qv49ictyFfxY=
 github.com/leodido/go-urn v1.2.4 h1:XlAE/cm/ms7TE/VMVoduSpNBoyc2dOxHs5MZSwAN63Q=
 github.com/leodido/go-urn v1.2.4/go.mod h1:7ZrI8mTSeBSHl/UaRyKQW1qZeMgak41ANeCNaVckg+4=
-github.com/lxn/walk v0.0.0-20210112085537-c389da54e794/go.mod h1:E23UucZGqpuUANJooIbHWCufXvOcT6E7Stq81gU+CSQ=
-github.com/lxn/win v0.0.0-20210218163916-a377121e959e/go.mod h1:KxxjdtRkfNoYDCUP5ryK7XJJNTnpC8atvtmTheChOtk=
 github.com/mattn/go-isatty v0.0.14/go.mod h1:7GGIvUiUoEMVVmxf/4nioHXj79iQHKdU27kJ6hsGG94=
 github.com/mattn/go-isatty v0.0.19 h1:JITubQf0MOLdlGRuRq+jtsDlekdYPia9ZFsB8h/APPA=
 github.com/mattn/go-isatty v0.0.19/go.mod h1:W+V8PltTTMOvKvAeJH7IuucS94S2C6jfK/D7dTCTo3Y=
@ -94,12 +75,8 @@ github.com/modern-go/reflect2 v1.0.2 h1:xBagoLtFs94CBntxluKeaWgTMpvLxC4ur3nMaC9G
 github.com/modern-go/reflect2 v1.0.2/go.mod h1:yWuevngMOJpCy52FWWMvUC8ws7m/LJsjYzDa0/r8luk=
 github.com/olekukonko/tablewriter v0.0.5 h1:P2Ga83D34wi1o9J6Wh1mRuqd4mF/x/lgBS7N7AbDhec=
 github.com/olekukonko/tablewriter v0.0.5/go.mod h1:hPp6KlRPjbx+hW8ykQs1w3UBbZlj6HuIJcUGPhkA7kY=
-github.com/oxtoacart/bpool v0.0.0-20190530202638-03653db5a59c h1:rp5dCmg/yLR3mgFuSOe4oEnDDmGLROTvMragMUXpTQw=
-github.com/oxtoacart/bpool v0.0.0-20190530202638-03653db5a59c/go.mod h1:X07ZCGwUbLaax7L0S3Tw4hpejzu63ZrrQiUe6W0hcy0=
 github.com/pbnjay/memory v0.0.0-20210728143218-7b4eea64cf58 h1:onHthvaw9LFnH4t2DcNVpwGmV9E1BkGknEliJkfwQj0=
 github.com/pbnjay/memory v0.0.0-20210728143218-7b4eea64cf58/go.mod h1:DXv8WO4yhMYhSNPKjeNKa5WY9YCIEBRbNzFFPJbWO6Y=
-github.com/pborman/uuid v1.2.1 h1:+ZZIw58t/ozdjRaXh/3awHfmWRbzYxJoAdNJxe/3pvw=
-github.com/pborman/uuid v1.2.1/go.mod h1:X/NO0urCmaxf9VXbdlT7C2Yzkj2IKimNn4k+gtPdI/k=
 github.com/pelletier/go-toml/v2 v2.0.1/go.mod h1:r9LEWfGN8R5k0VXJ+0BkIe7MYkRdwZOjgMj2KwnJFUo=
 github.com/pelletier/go-toml/v2 v2.0.8 h1:0ctb6s9mE31h0/lhu+J6OPmVeDxJn+kYnJc2jZR9tGQ=
 github.com/pelletier/go-toml/v2 v2.0.8/go.mod h1:vuYfssBdrU2XDZ9bYydBu6t+6a6PYNcZljzZR9VXg+4=
@ -112,7 +89,6 @@ github.com/rogpeppe/go-internal v1.6.1/go.mod h1:xXDCJY+GAPziupqXw64V24skbSoqbTE
 github.com/rogpeppe/go-internal v1.8.0 h1:FCbCCtXNOY3UtUuHUYaghJg4y7Fd14rXifAYUAtL9R8=
 github.com/rogpeppe/go-internal v1.8.0/go.mod h1:WmiCO8CzOY8rg0OYDC4/i/2WRWAB6poM+XZ2dLUbcbE=
 github.com/russross/blackfriday/v2 v2.1.0/go.mod h1:+Rmxgy9KzJVeS9/2gXHxylqXiyQDYRxCVz55jmeOWTM=
-github.com/skratchdot/open-golang v0.0.0-20200116055534-eef842397966/go.mod h1:sUM3LWHvSMaG192sy56D9F7CNvL7jUJVXoqM1QKLnog=
 github.com/spf13/cobra v1.7.0 h1:hyqWnYt1ZQShIddO5kBpj3vu05/++x6tJ6dg8EC572I=
 github.com/spf13/cobra v1.7.0/go.mod h1:uLxZILRyS/50WlhOIKD7W6V5bgeIt+4sICxh6uRMrb0=
 github.com/spf13/pflag v1.0.5 h1:iy+VFUOCP1a+8yFto/drg2CJ5u0yRoB7fZw3DKv/JXA=
@ -149,14 +125,12 @@ golang.org/x/net v0.17.0 h1:pVaXccu2ozPjCXewfr1S7xza/zcXTity9cCdXQYSjIM=
 golang.org/x/net v0.17.0/go.mod h1:NxSsAGuq816PNPmqtQdLE42eU2Fs7NoRIZrHJAlaCOE=
 golang.org/x/sync v0.3.0 h1:ftCYgMx6zT/asHUrPw8BLLscYtGznsLAnjq5RH9P66E=
 golang.org/x/sync v0.3.0/go.mod h1:FU7BRWz2tNW+3quACPkgCx/L+uEAv1htQ0V83Z9Rj+Y=
-golang.org/x/sys v0.0.0-20201018230417-eeed37f84f13/go.mod h1:h1NjWce9XRLGQEsW7wpKNCjG9DtNlClVuFLEZdDNbEs=
 golang.org/x/sys v0.0.0-20201119102817-f84b799fce68/go.mod h1:h1NjWce9XRLGQEsW7wpKNCjG9DtNlClVuFLEZdDNbEs=
 golang.org/x/sys v0.0.0-20210124154548-22da62e12c0c/go.mod h1:h1NjWce9XRLGQEsW7wpKNCjG9DtNlClVuFLEZdDNbEs=
 golang.org/x/sys v0.0.0-20210615035016-665e8c7367d1/go.mod h1:oPkhp1MJrh7nUepCBck5+mAzfO9JrbApNNgaTdGDITg=
 golang.org/x/sys v0.0.0-20210630005230-0f9fa26af87c/go.mod h1:oPkhp1MJrh7nUepCBck5+mAzfO9JrbApNNgaTdGDITg=
 golang.org/x/sys v0.0.0-20210806184541-e5e7981a1069/go.mod h1:oPkhp1MJrh7nUepCBck5+mAzfO9JrbApNNgaTdGDITg=
 golang.org/x/sys v0.0.0-20220704084225-05e143d24a9e/go.mod h1:oPkhp1MJrh7nUepCBck5+mAzfO9JrbApNNgaTdGDITg=
-golang.org/x/sys v0.1.0/go.mod h1:oPkhp1MJrh7nUepCBck5+mAzfO9JrbApNNgaTdGDITg=
 golang.org/x/sys v0.6.0/go.mod h1:oPkhp1MJrh7nUepCBck5+mAzfO9JrbApNNgaTdGDITg=
 golang.org/x/sys v0.13.0 h1:Af8nKPmuFypiUBjVoU9V20FiaFXOcuZI21p0ycVYYGE=
 golang.org/x/sys v0.13.0/go.mod h1:oPkhp1MJrh7nUepCBck5+mAzfO9JrbApNNgaTdGDITg=
@ -173,7 +147,6 @@ google.golang.org/protobuf v1.26.0-rc.1/go.mod h1:jlhhOSvTdKEhbULTjvd4ARK9grFBp0
 google.golang.org/protobuf v1.28.0/go.mod h1:HV8QOd/L58Z+nl8r43ehVNZIU/HEI6OcFqwMG9pJV4I=
 google.golang.org/protobuf v1.30.0 h1:kPPoIgf3TsEvrm0PFe15JQ+570QVxYzEvvHqChK+cng=
 google.golang.org/protobuf v1.30.0/go.mod h1:HV8QOd/L58Z+nl8r43ehVNZIU/HEI6OcFqwMG9pJV4I=
-gopkg.in/Knetic/govaluate.v3 v3.0.0/go.mod h1:csKLBORsPbafmSCGTEh3U7Ozmsuq8ZSIlKk1bcqph0E=
 gopkg.in/check.v1 v0.0.0-20161208181325-20d25e280405/go.mod h1:Co6ibVJAznAaIkqp8huTwlJQCZ016jof/cbN4VW5Yz0=
 gopkg.in/check.v1 v1.0.0-20180628173108-788fd7840127/go.mod h1:Co6ibVJAznAaIkqp8huTwlJQCZ016jof/cbN4VW5Yz0=
 gopkg.in/check.v1 v1.0.0-20201130134442-10cb98267c6c h1:Hei/4ADfdWqJk1ZMxUNpqntNwaWcugrBjAiHlqqRiVk=
--- a/gpu/gpu_darwin.go
+++ b/gpu/gpu_darwin.go
@ -1,12 +1,14 @@
 //go:build darwin

 package gpu
-
+/*
+#cgo CFLAGS: -x objective-c
+#cgo LDFLAGS: -framework Foundation -framework CoreGraphics -framework Metal
+#include "gpu_info_darwin.h"
+*/
 import "C"
 import (
 	"runtime"
-
-	"github.com/pbnjay/memory"
 )

 // CheckVRAM returns the free VRAM in bytes on Linux machines with NVIDIA GPUs
@ -15,19 +17,8 @@ func CheckVRAM() (int64, error) {
 		// gpu not supported, this may not be metal
 		return 0, nil
 	}
-
-	// on macOS, there's already buffer for available vram (see below) so just return the total
-	systemMemory := int64(memory.TotalMemory())
-
-	// macOS limits how much memory is available to the GPU based on the amount of system memory
-	// TODO: handle case where iogpu.wired_limit_mb is set to a higher value
-	if systemMemory <= 36*1024*1024*1024 {
-		systemMemory = systemMemory * 2 / 3
-	} else {
-		systemMemory = systemMemory * 3 / 4
-	}
-
-	return systemMemory, nil
+	recommendedMaxVRAM := int64(C.getRecommendedMaxVRAM())
+	return recommendedMaxVRAM, nil
 }

 func GetGPUInfo() GpuInfo {
--- a/gpu/gpu_info_cuda.c
+++ b/gpu/gpu_info_cuda.c
@ -156,7 +156,7 @@ void cuda_check_vram(cuda_handle_t h, mem_info_t *resp) {
    }

    LOG(h.verbose, "[%d] CUDA totalMem %ld\n", i, memInfo.total);
-    LOG(h.verbose, "[%d] CUDA usedMem %ld\n", i, memInfo.free);
+    LOG(h.verbose, "[%d] CUDA usedMem %ld\n", i, memInfo.used);

    resp->total += memInfo.total;
    resp->free += memInfo.free;
--- a/gpu/gpu_info_darwin.h
+++ b/gpu/gpu_info_darwin.h
@ -0,0 +1,3 @@
+#import <Metal/Metal.h>
+#include <stdint.h>
+uint64_t getRecommendedMaxVRAM();
--- a/gpu/gpu_info_darwin.m
+++ b/gpu/gpu_info_darwin.m
@ -0,0 +1,11 @@
+//go:build darwin
+#include "gpu_info_darwin.h"
+
+uint64_t getRecommendedMaxVRAM()
+{
+	id<MTLDevice> device = MTLCreateSystemDefaultDevice();
+	uint64_t result = device.recommendedMaxWorkingSetSize;
+	CFRelease(device);
+	return result;
+}
+
--- a/llm/ext_server/ext_server.cpp
+++ b/llm/ext_server/ext_server.cpp
@ -146,9 +146,9 @@ void llama_server_start() {
      llama->queue_tasks.on_new_task(std::bind(
        &llama_server_context::process_single_task, llama, std::placeholders::_1));
      llama->queue_tasks.on_finish_multitask(std::bind(
-          &llama_server_context::on_finish_multitask, llama, std::placeholders::_1));
-      llama->queue_tasks.on_all_tasks_finished(std::bind(
-          &llama_server_context::run_on_all_tasks_finished, llama));
+        &llama_server_context::on_finish_multitask, llama, std::placeholders::_1));
+      llama->queue_tasks.on_run_slots(std::bind(
+        &llama_server_context::update_slots, llama));
      llama->queue_results.on_multitask_update(std::bind(
          &llama_server_queue::update_multitask,
          &llama->queue_tasks,
--- a/llm/generate/gen_common.sh
+++ b/llm/generate/gen_common.sh
@ -101,7 +101,7 @@ compress_libs() {
    pids=""
    rm -rf ${BUILD_DIR}/lib/*.${LIB_EXT}*.gz
    for lib in ${BUILD_DIR}/lib/*.${LIB_EXT}* ; do
-        gzip --best -f ${lib} &
+        gzip -n --best -f ${lib} &
        pids+=" $!"
    done
    echo 
--- a/llm/llama.cpp
+++ b/llm/llama.cpp
--- a/llm/patches/01-cache.diff
+++ b/llm/patches/01-cache.diff
@ -1,9 +1,9 @@
 diff --git a/examples/server/server.cpp b/examples/server/server.cpp
-index d86d7e04..2694e92e 100644
+index 2b2f4a0f..afac49af 100644
 --- a/examples/server/server.cpp
 +++ b/examples/server/server.cpp
-@@ -901,13 +901,15 @@ struct llama_server_context
-                 slot.sent_count += result.text_to_send.size();
+@@ -997,13 +997,15 @@ struct llama_server_context
+                 slot.n_sent_text += result.text_to_send.size();
                 // add the token to slot queue and cache
             }
 -            slot.add_token_string(result);
--- a/llm/patches/02-cudaleaks.diff
+++ b/llm/patches/02-cudaleaks.diff
@ -1,8 +1,8 @@
 diff --git a/examples/server/server.cpp b/examples/server/server.cpp
-index 7800c6e7..be30db23 100644
+index 2b2f4a0f..25857bdd 100644
 --- a/examples/server/server.cpp
 +++ b/examples/server/server.cpp
-@@ -30,6 +30,10 @@
+@@ -31,6 +31,10 @@
 #include <atomic>
 #include <signal.h>
 
@ -12,8 +12,8 @@ index 7800c6e7..be30db23 100644
 +
 using json = nlohmann::json;
 
- struct server_params
-@@ -353,6 +357,9 @@ struct llama_server_context
+ struct server_params {
+@@ -363,6 +367,9 @@ struct llama_server_context
             llama_free_model(model);
             model = nullptr;
         }
@ -23,7 +23,7 @@ index 7800c6e7..be30db23 100644
     }
 
     bool load_model(const gpt_params &params_)
-@@ -3143,6 +3150,7 @@ int main(int argc, char **argv)
+@@ -3494,6 +3501,7 @@ int main(int argc, char **argv)
     sigemptyset (&sigint_action.sa_mask);
     sigint_action.sa_flags = 0;
     sigaction(SIGINT, &sigint_action, NULL);
@ -32,10 +32,10 @@ index 7800c6e7..be30db23 100644
     auto console_ctrl_handler = +[](DWORD ctrl_type) -> BOOL {
         return (ctrl_type == CTRL_C_EVENT) ? (signal_handler(SIGINT), true) : false;
 diff --git a/ggml-cuda.cu b/ggml-cuda.cu
-index 933ebbc4..88a4f664 100644
+index 0c6501e9..75c12723 100644
 --- a/ggml-cuda.cu
 +++ b/ggml-cuda.cu
-@@ -39,6 +39,7 @@
+@@ -43,6 +43,7 @@
 #define __shfl_xor_sync(mask, var, laneMask, width) __shfl_xor(var, laneMask, width)
 #define cublasComputeType_t hipblasDatatype_t //deprecated, new hipblasComputeType_t not in 5.6
 #define cublasCreate hipblasCreate
@ -43,7 +43,7 @@ index 933ebbc4..88a4f664 100644
 #define cublasGemmEx hipblasGemmEx
 #define cublasGemmBatchedEx hipblasGemmBatchedEx
 #define cublasGemmStridedBatchedEx hipblasGemmStridedBatchedEx
-@@ -7991,10 +7992,10 @@ GGML_CALL bool ggml_cublas_loaded(void) {
+@@ -8694,10 +8695,10 @@ GGML_CALL bool ggml_cublas_loaded(void) {
     return g_cublas_loaded;
 }
 
@ -57,7 +57,7 @@ index 933ebbc4..88a4f664 100644
 
 #ifdef __HIP_PLATFORM_AMD__
         // Workaround for a rocBLAS bug when using multiple graphics cards:
-@@ -8004,7 +8005,7 @@ GGML_CALL void ggml_init_cublas() {
+@@ -8707,7 +8708,7 @@ GGML_CALL void ggml_init_cublas() {
 #endif
 
         if (cudaGetDeviceCount(&g_device_count) != cudaSuccess) {
@ -66,7 +66,7 @@ index 933ebbc4..88a4f664 100644
             g_cublas_loaded = false;
             fprintf(stderr, "%s: no " GGML_CUDA_NAME " devices found, " GGML_CUDA_NAME " will be disabled\n", __func__);
             return;
-@@ -8075,7 +8076,7 @@ GGML_CALL void ggml_init_cublas() {
+@@ -8778,7 +8779,7 @@ GGML_CALL void ggml_init_cublas() {
         // configure logging to stdout
         // CUBLAS_CHECK(cublasLoggerConfigure(1, 1, 0, nullptr));
 
@ -75,12 +75,11 @@ index 933ebbc4..88a4f664 100644
         g_cublas_loaded = true;
     }
 }
-@@ -11604,3 +11605,23 @@ GGML_CALL int ggml_backend_cuda_reg_devices() {
+@@ -12345,3 +12346,22 @@ GGML_CALL int ggml_backend_cuda_reg_devices() {
     }
     return device_count;
 }
 +
-+
 +extern "C" GGML_CALL void ggml_free_cublas(void);
 +GGML_CALL void ggml_free_cublas(void) {
 +    for (int id = 0; id < g_device_count; ++id) {
@ -100,16 +99,16 @@ index 933ebbc4..88a4f664 100644
 +    g_cublas_initialized = false;
 +}
 diff --git a/ggml-cuda.h b/ggml-cuda.h
-index b1ebd61d..b4c80c2c 100644
+index b1ebd61d..6dd58ddf 100644
 --- a/ggml-cuda.h
 +++ b/ggml-cuda.h
-@@ -20,6 +20,9 @@ extern "C" {
- // Always success. To check if CUDA is actually loaded, use `ggml_cublas_loaded`.
- GGML_API GGML_CALL void   ggml_init_cublas(void);
+@@ -23,6 +23,9 @@ GGML_API GGML_CALL void   ggml_init_cublas(void);
+ // Returns `true` if there are available CUDA devices and cublas loads successfully; otherwise, it returns `false`.
+ GGML_API GGML_CALL bool   ggml_cublas_loaded(void);
 
 +// Release CUDA resources
 +GGML_API GGML_CALL void   ggml_free_cublas(void);
 +
- // Returns `true` if there are available CUDA devices and cublas loads successfully; otherwise, it returns `false`.
- GGML_API GGML_CALL bool   ggml_cublas_loaded(void);
+ GGML_API GGML_CALL void * ggml_cuda_host_malloc(size_t size);
+ GGML_API GGML_CALL void   ggml_cuda_host_free(void * ptr);
 
--- a/scripts/build_docker.sh
+++ b/scripts/build_docker.sh
@ -5,13 +5,15 @@ set -eu
 export VERSION=${VERSION:-$(git describe --tags --first-parent --abbrev=7 --long --dirty --always | sed -e "s/^v//g")}
 export GOFLAGS="'-ldflags=-w -s \"-X=github.com/jmorganca/ollama/version.Version=$VERSION\" \"-X=github.com/jmorganca/ollama/server.mode=release\"'"

+IMAGE_NAME=${IMAGE_NAME:-"ollama/ollama"}
+BUILD_PLATFORM=${BUILD_PLATFORM:-"linux/arm64,linux/amd64"}
 docker build \
    --load \
-    --platform=linux/arm64,linux/amd64 \
+    --platform=${BUILD_PLATFORM} \
    --build-arg=VERSION \
    --build-arg=GOFLAGS \
    -f Dockerfile \
-    -t ollama/ollama:$VERSION \
+    -t ${IMAGE_NAME}:$VERSION \
    .

 docker build \
@ -21,5 +23,12 @@ docker build \
    --build-arg=GOFLAGS \
    --target runtime-rocm \
    -f Dockerfile \
-    -t ollama/ollama:$VERSION-rocm \
+    -t ${IMAGE_NAME}:$VERSION-rocm \
    .
+
+docker tag ${IMAGE_NAME}:$VERSION ${IMAGE_NAME}:latest
+docker tag ${IMAGE_NAME}:$VERSION-rocm ${IMAGE_NAME}:rocm
+
+echo "To release, run:"
+echo "  docker push ${IMAGE_NAME}:$VERSION && docker push ${IMAGE_NAME}:latest"
+echo "  docker push ${IMAGE_NAME}:$VERSION-rocm && docker push ${IMAGE_NAME}:rocm"
--- a/scripts/build_windows.ps1
+++ b/scripts/build_windows.ps1
@ -53,7 +53,7 @@ function buildOllama() {
    write-host "Building ollama CLI"
    & go generate ./...
    if ($LASTEXITCODE -ne 0) { exit($LASTEXITCODE)}
-    & go build -ldflags "-X=github.com/jmorganca/ollama/version.Version=$script:VERSION -X=github.com/jmorganca/ollama/server.mode=release" .
+    & go build -ldflags "-s -w -X=github.com/jmorganca/ollama/version.Version=$script:VERSION -X=github.com/jmorganca/ollama/server.mode=release" .
    if ($LASTEXITCODE -ne 0) { exit($LASTEXITCODE)}
    if ("${env:KEY_CONTAINER}") {
        & "${script:SignTool}" sign /v /fd sha256 /t http://timestamp.digicert.com /f "${script:OLLAMA_CERT}" `
@ -68,7 +68,7 @@ function buildApp() {
    write-host "Building Ollama App"
    cd "${script:SRC_DIR}\app"
    & windres -l 0 -o ollama.syso ollama.rc
-    & go build -ldflags "-H windowsgui -X=github.com/jmorganca/ollama/version.Version=$script:VERSION -X=github.com/jmorganca/ollama/server.mode=release" .
+    & go build -ldflags "-s -w -H windowsgui -X=github.com/jmorganca/ollama/version.Version=$script:VERSION -X=github.com/jmorganca/ollama/server.mode=release" .
    if ($LASTEXITCODE -ne 0) { exit($LASTEXITCODE)}
    if ("${env:KEY_CONTAINER}") {
        & "${script:SignTool}" sign /v /fd sha256 /t http://timestamp.digicert.com /f "${script:OLLAMA_CERT}" `
--- a/scripts/install.sh
+++ b/scripts/install.sh
@ -88,6 +88,10 @@ configure_systemd() {
        status "Adding ollama user to render group..."
        $SUDO usermod -a -G render ollama
    fi
+    if getent group video >/dev/null 2>&1; then
+        status "Adding ollama user to video group..."
+        $SUDO usermod -a -G video ollama
+    fi

    status "Adding current user to ollama group..."
    $SUDO usermod -a -G ollama $(whoami)
--- a/server/prompt.go
+++ b/server/prompt.go
@ -121,13 +121,15 @@ func ChatPrompt(tmpl string, messages []api.Message, window int, encode func(str
 				p = prompt{}
 			}

-			p.Prompt = msg.Content
-
+			var sb strings.Builder
 			for range msg.Images {
-				p.Prompt += fmt.Sprintf(" [img-%d]", imgId)
+				fmt.Fprintf(&sb, "[img-%d] ", imgId)
 				p.images = append(p.images, imgId)
 				imgId += 1
 			}
+
+			sb.WriteString(msg.Content)
+			p.Prompt = sb.String()
 		case "assistant":
 			if p.Response != "" {
 				prompts = append(prompts, p)
--- a/server/prompt_test.go
+++ b/server/prompt_test.go
@ -155,7 +155,7 @@ func TestChatPrompt(t *testing.T) {
 				{Role: "user", Content: "Hello", Images: []api.ImageData{[]byte("base64")}},
 			},
 			window: 1024,
-			want:   "You are a Wizard. Hello [img-0]",
+			want:   "You are a Wizard. [img-0] Hello",
 		},
 		{
 			name:     "images truncated",
@ -165,7 +165,7 @@ func TestChatPrompt(t *testing.T) {
 				{Role: "user", Content: "Hello", Images: []api.ImageData{[]byte("img1"), []byte("img2")}},
 			},
 			window: 1024,
-			want:   "You are a Wizard. Hello [img-1]",
+			want:   "You are a Wizard. [img-0] [img-1] Hello",
 		},
 		{
 			name:     "empty list",
@ -198,7 +198,7 @@ func TestChatPrompt(t *testing.T) {
 			}

 			if got != tc.want {
-				t.Errorf("got = %v, want %v", got, tc.want)
+				t.Errorf("got: %q, want: %q", got, tc.want)
 			}
 		})
 	}
--- a/server/routes.go
+++ b/server/routes.go
@ -250,6 +250,19 @@ func GenerateHandler(c *gin.Context) {
 		slog.Debug("generate handler", "system", req.System)

 		var sb strings.Builder
+		for i := range req.Images {
+			fmt.Fprintf(&sb, "[img-%d] ", i)
+		}
+
+		sb.WriteString(req.Prompt)
+
+		p, err := Prompt(req.Template, req.System, sb.String(), "", true)
+		if err != nil {
+			c.JSON(http.StatusInternalServerError, gin.H{"error": err.Error()})
+			return
+		}
+
+		sb.Reset()
 		if req.Context != nil {
 			prev, err := loaded.runner.Decode(c.Request.Context(), req.Context)
 			if err != nil {
@ -260,18 +273,6 @@ func GenerateHandler(c *gin.Context) {
 			sb.WriteString(prev)
 		}

-		// write image tags
-		// TODO: limit the number of images to fit in the context similar to the chat endpoint
-		for i := range req.Images {
-			req.Prompt += fmt.Sprintf(" [img-%d]", i)
-		}
-
-		p, err := Prompt(req.Template, req.System, req.Prompt, "", true)
-		if err != nil {
-			c.JSON(http.StatusInternalServerError, gin.H{"error": err.Error()})
-			return
-		}
-
 		sb.WriteString(p)

 		prompt = sb.String()
@ -384,7 +385,7 @@ func GenerateHandler(c *gin.Context) {
 	streamResponse(c, ch)
 }

-func EmbeddingHandler(c *gin.Context) {
+func EmbeddingsHandler(c *gin.Context) {
 	loaded.mu.Lock()
 	defer loaded.mu.Unlock()

@ -437,8 +438,9 @@ func EmbeddingHandler(c *gin.Context) {
 		return
 	}

-	if !loaded.Options.EmbeddingOnly {
-		c.JSON(http.StatusBadRequest, gin.H{"error": "embedding option must be set to true"})
+	// an empty request loads the model
+	if req.Prompt == "" {
+		c.JSON(http.StatusOK, api.EmbeddingResponse{Embedding: []float64{}})
 		return
 	}

@ -947,7 +949,7 @@ func (s *Server) GenerateRoutes() http.Handler {
 	r.POST("/api/pull", PullModelHandler)
 	r.POST("/api/generate", GenerateHandler)
 	r.POST("/api/chat", ChatHandler)
-	r.POST("/api/embeddings", EmbeddingHandler)
+	r.POST("/api/embeddings", EmbeddingsHandler)
 	r.POST("/api/create", CreateModelHandler)
 	r.POST("/api/push", PushModelHandler)
 	r.POST("/api/copy", CopyModelHandler)
Author	SHA1	Message	Date
Jeffrey Morgan	21347e1ed6	update llama.cpp submodule to `c29af7e` (#2868 )	2024-03-01 15:26:04 -08:00
Jeffrey Morgan	3b4bab3dc5	Fix embeddings load model behavior (#2848 )	2024-02-29 17:40:56 -08:00
Daniel Hiltgen	cbd6e3b38e	Merge pull request #2838 from dhiltgen/opensuse Add ollama user to video group	2024-02-29 15:47:56 -08:00
Daniel Hiltgen	b830afa716	Merge pull request #2837 from dhiltgen/podman_image_support Add env var so podman will map cuda GPUs	2024-02-29 15:47:37 -08:00
Daniel Hiltgen	bd1d8b0d14	Merge pull request #2836 from bmwiedemann/gzip Omit build date from gzip headers	2024-02-29 15:46:46 -08:00
fred-bf	25c2912120	Add Community Integration: NextChat (#2780 )	2024-02-29 12:12:13 -08:00
Michael Yang	0e19476b56	prepend image tags (#2789 ) instead of appending image tags, prepend them - this generally produces better results	2024-02-29 11:30:14 -08:00
tylinux	fa2f2b3563	fix: print usedMemory size right (#2827 )	2024-02-29 11:11:04 -08:00
Jeffrey Morgan	cbf4970e0f	bump submodule to `87c91c07663b707e831c59ec373b5e665ff9d64a` (#2828 )	2024-02-29 09:42:08 -08:00
Daniel Hiltgen	74468513bd	Add ollama user to video group On OpenSUSE, ollama needs to be a member of the video group to access the GPU	2024-02-29 08:50:10 -08:00
Daniel Hiltgen	794a916a72	Add env var so podman will map cuda GPUs Without this env var, podman's GPU logic doesn't map the GPU through	2024-02-29 08:43:08 -08:00
Bernhard M. Wiedemann	76e5d9ec88	Omit build date from gzip headers See https://reproducible-builds.org/ for why this is good. This patch was done while working on reproducible builds for openSUSE.	2024-02-29 16:48:19 +01:00
Daniel Hiltgen	076237b8ea	Merge pull request #2771 from dhiltgen/toggle_models Bump llama.cpp to b2276	2024-02-27 11:29:53 -08:00
Daniel Hiltgen	53d694c67f	Merge pull request #2772 from dhiltgen/container_image Refine container image build script	2024-02-27 11:29:08 -08:00
Daniel Hiltgen	5aa6bfea94	Merge pull request #2785 from dhiltgen/win_download Log unexpected server errors checking for update	2024-02-27 10:43:14 -08:00
Daniel Hiltgen	1cde63dd64	Log unexpected server errors checking for update This should unmask some failure modes that likely show up in app logs as unmarshal errors	2024-02-27 09:17:04 -08:00
Daniel Hiltgen	98e0b7e94f	Refine container image build script Allow overriding the platform, image name, and tag latest for standard and rocm images.	2024-02-26 17:26:49 -08:00
Daniel Hiltgen	061e8f6abc	Bump llama.cpp to b2276	2024-02-26 16:49:24 -08:00
peanut256	a189810df6	Determine max VRAM on macOS using `recommendedMaxWorkingSetSize` (#2354 ) * read iogpu.wired_limit_mb on macOS Fix for https://github.com/ollama/ollama/issues/1826 * improved determination of available vram on macOS read the recommended maximal vram on macOS via Metal API * Removed macOS-specific logging * Remove logging from gpu_darwin.go * release Core Foundation object fixes a possible memory leak	2024-02-25 18:16:45 -05:00
Ikko Eltociear Ashimine	e95b896790	Update types.go (#2744 ) specfied -> specified	2024-02-25 13:41:25 -05:00
elthommy	1f087c4d26	Update langchain python tutorial (#2737 ) Remove unused GPT4all Use nomic-embed-text as embedded model Fix a deprecation warning (__call__)	2024-02-25 00:31:36 -05:00
Jeffrey Morgan	5d7ea6616f	no extra disk space for windows installation (#2739 )	2024-02-25 00:20:35 -05:00
Michael Yang	2a4b128ae3	Merge pull request #2719 from ollama/mxyng/format-private-key remove format private key	2024-02-23 17:15:14 -08:00
Michael Yang	fc483274ad	clean up go.mod	2024-02-23 16:53:36 -08:00
Michael Yang	fd10a2ad4b	remove format/openssh.go this is unnecessary now that x/crypto/ssh.MarshalPrivateKey has been added	2024-02-23 16:52:23 -08:00
Benn Huang	b291f63188	Add Community Integration: Chatbox Co-authored-by: bennhuang <bennhuang@tencent.com>	2024-02-23 07:17:28 -05:00
Jeffrey Morgan	f58856bf6f	better directory cleanup in `ollama.iss`	2024-02-23 07:14:59 -05:00
Jeffrey Morgan	275ea01587	restore windows build flags and compression	2024-02-22 18:07:18 -05:00