fix conversion

convert/convert.go

@@ -190,8 +190,8 @@ func ConvertModel(fsys fs.FS, ws io.WriteSeeker) error {
 		conv = &gemmaModel{}
 	case "Gemma2ForCausalLM":
 		conv = &gemma2Model{}
-	case "Gemma3ForConditionalGeneration":
-		conv = &gemma3Model{}
+	case "Gemma3ForCausalLM", "Gemma3ForConditionalGeneration":
+		conv = &gemma3Model{Architecture: p.Architectures[0]}
 	case "Phi3ForCausalLM":
 		conv = &phi3Model{}
 	case "Qwen2ForCausalLM":
@@ -226,6 +226,9 @@ func ConvertModel(fsys fs.FS, ws io.WriteSeeker) error {
 	}
 
 	switch {
+	case vocabSize == 0:
+		slog.Warn("vocabulary size was not explicitly set by the model", "default size", len(t.Vocabulary.Tokens))
+		vocabSize = len(t.Vocabulary.Tokens)
 	case vocabSize > len(t.Vocabulary.Tokens):
 		slog.Warn("vocabulary is smaller than expected, padding with dummy tokens", "expect", vocabSize, "actual", len(t.Vocabulary.Tokens))
 		for i := range vocabSize - len(t.Vocabulary.Tokens) {

convert/convert_gemma3.go

@@ -4,7 +4,13 @@ import "github.com/ollama/ollama/fs/ggml"
 
 type gemma3Model struct {
 	gemmaModel
-	TextModel   gemma3TextModel `json:"text_config"`
+	Architecture string
+	TextModel    struct {
+		HiddenSize       uint32 `json:"hidden_size"`
+		HiddenLayers     uint32 `json:"num_hidden_layers"`
+		IntermediateSize uint32 `json:"intermediate_size"`
+		SlidingWindow    uint32 `json:"sliding_window"`
+	} `json:"text_config"`
 	VisionModel struct {
 		NumAttentionHeads uint32  `json:"num_attention_heads"` // attention.head_count 16
 		LayerNormEpsilon  float32 `json:"layer_norm_eps"`      // attention.layer_norm_epsilon 1e-05
@@ -15,41 +21,41 @@ type gemma3Model struct {
 		NumChannels       uint32  `json:"num_channels"`        // num_channels 3
 		PatchSize         uint32  `json:"patch_size"`          // patch_size 14
 	} `json:"vision_config"`
-}
-
-type gemma3TextModel struct {
 	MaxPositionEmbeddings uint32  `json:"max_position_embeddings"`
-	HiddenSize            uint32  `json:"hidden_size"`
-	HiddenLayers          uint32  `json:"num_hidden_layers"`
-	IntermediateSize      uint32  `json:"intermediate_size"`
 	NumAttentionHeads     uint32  `json:"num_attention_heads"`
 	NumKeyValueHeads      uint32  `json:"num_key_value_heads"`
 	RMSNormEPS            float32 `json:"rms_norm_eps"`
 	HeadDim               uint32  `json:"head_dim"`
-	SlidingWindow         uint32  `json:"sliding_window"`
-	AttentionLogitSoftcap float32 `json:"attn_logit_softcapping"`
 	FinalLogitSoftcap     float32 `json:"final_logit_softcapping"`
 	RopeLocalTheta        float32 `json:"rope_local_base_freq"`
 	RopeGlobalTheta       float32 `json:"rope_global_base_freq"`
+	SlidingWindow         uint32  `json:"sliding_window"`
 }
 
 func (p *gemma3Model) KV(t *Tokenizer) ggml.KV {
 	kv := p.ModelParameters.KV(t)
 	kv["general.architecture"] = "gemma3"
-	kv["gemma3.context_length"] = p.TextModel.MaxPositionEmbeddings
+
+	switch p.Architecture {
+	case "Gemma3ForCausalLM":
+		kv["gemma3.context_length"] = p.MaxPositionEmbeddings
+		kv["gemma3.attention.head_count"] = p.NumAttentionHeads
+		kv["gemma3.attention.head_count_kv"] = p.NumKeyValueHeads
+		kv["gemma3.text.attention.layer_norm_rms_epsilon"] = p.RMSNormEPS
+		kv["gemma3.attention.key_length"] = p.HeadDim
+		kv["gemma3.attention.value_length"] = p.HeadDim
+		kv["gemma3.text.attention.sliding_window"] = p.SlidingWindow
+		kv["gemma3.text.final_logit_softcapping"] = p.FinalLogitSoftcap
+		kv["gemma3.text.rope.local.freq_base"] = p.RopeLocalTheta
+		kv["gemma3.text.rope.global.freq_base"] = p.RopeGlobalTheta
+		kv["gemma3.embedding_length"] = p.HiddenSize
+		kv["gemma3.block_count"] = p.HiddenLayers
+		kv["gemma3.text.feed_forward_length"] = p.IntermediateSize
+	default:
 		kv["gemma3.embedding_length"] = p.TextModel.HiddenSize
 		kv["gemma3.block_count"] = p.TextModel.HiddenLayers
 		kv["gemma3.text.feed_forward_length"] = p.TextModel.IntermediateSize
-	kv["gemma3.attention.head_count"] = p.TextModel.NumAttentionHeads
-	kv["gemma3.attention.head_count_kv"] = p.TextModel.NumKeyValueHeads
-	kv["gemma3.text.attention.layer_norm_rms_epsilon"] = p.TextModel.RMSNormEPS
-	kv["gemma3.attention.key_length"] = p.TextModel.HeadDim
-	kv["gemma3.attention.value_length"] = p.TextModel.HeadDim
 		kv["gemma3.text.attention.sliding_window"] = p.TextModel.SlidingWindow
-	kv["gemma3.text.final_logit_softcapping"] = p.TextModel.FinalLogitSoftcap
-	kv["gemma3.text.rope.local.freq_base"] = p.TextModel.RopeLocalTheta
-	kv["gemma3.text.rope.global.freq_base"] = p.TextModel.RopeGlobalTheta
-
 		kv["gemma3.vision.block_count"] = p.VisionModel.NumHiddenLayers
 		kv["gemma3.vision.embedding_length"] = p.VisionModel.HiddenSize
 		kv["gemma3.vision.feed_forward_length"] = p.VisionModel.IntermediateSize
@@ -58,6 +64,11 @@ func (p *gemma3Model) KV(t *Tokenizer) ggml.KV {
 		kv["gemma3.vision.num_channels"] = p.VisionModel.NumChannels
 		kv["gemma3.vision.attention.head_count"] = p.VisionModel.NumAttentionHeads
 		kv["gemma3.vision.attention.layer_norm_epsilon"] = p.VisionModel.LayerNormEpsilon
+	}
+
+	kv["tokenizer.ggml.bos_token_id"] = uint32(2)
+	kv["tokenizer.ggml.eot_token_id"] = uint32(1)
+
 	return kv
 }
 

model/models/gemma3/model_text.go

@@ -32,6 +32,7 @@ type TextModel struct {
 }
 
 const (
+	gemmaGlobalCacheCount = 6
 	gemma27BLayerCount    = 46
 )
 
@@ -55,15 +56,15 @@ func newTextModel(c ml.Config) *TextModel {
 		Layers: make([]TextLayer, c.Uint("block_count")),
 		TextOptions: &TextOptions{
 			hiddenSize:        int(c.Uint("embedding_length")),
-			numHeads:          int(c.Uint("attention.head_count")),
-			numKVHeads:        int(c.Uint("attention.head_count_kv")),
-			attnKeyLen:        int(c.Uint("attention.key_length")),
-			attnValLen:        int(c.Uint("attention.value_length")),
-			eps:               c.Float("text.attention.layer_norm_rms_epsilon"),
+			numHeads:          int(c.Uint("attention.head_count", 8)),
+			numKVHeads:        int(c.Uint("attention.head_count_kv", 4)),
+			attnKeyLen:        int(c.Uint("attention.key_length", 256)),
+			attnValLen:        int(c.Uint("attention.value_length", 256)),
+			eps:               c.Float("text.attention.layer_norm_rms_epsilon", 1e-06),
 			ropeLocalBase:     c.Float("text.rope.local.freq_base", 10000.0),
 			ropeGlobalBase:    c.Float("text.rope.global.freq_base", 1000000.0),
 			ropeScale:         c.Float("text.rope.freq_scale", 1.0),
-			finalLogitSoftcap: c.Float("text.final_logit_softcapping"),
+			finalLogitSoftcap: c.Float("text.final_logit_softcapping", 30.0),
 		},
 	}
 
@@ -84,7 +85,7 @@ func (sa *TextSelfAttention) Forward(ctx ml.Context, layer int, hiddenState, pos
 	ropeType := uint32(2)
 
 	ropeBase := opts.ropeLocalBase
-	if (layer+1)%6 == 0 {
+	if (layer+1)%gemmaGlobalCacheCount == 0 {
 		ropeBase = opts.ropeGlobalBase
 	}
 
@@ -116,7 +117,7 @@ func (sa *TextSelfAttention) Forward(ctx ml.Context, layer int, hiddenState, pos
 
 func (m *TextModel) Shift(ctx ml.Context, layer int, key, shift ml.Tensor) (ml.Tensor, error) {
 	ropeBase := m.TextOptions.ropeLocalBase
-	if (layer+1)%6 == 0 {
+	if (layer+1)%gemmaGlobalCacheCount == 0 {
 		ropeBase = m.TextOptions.ropeGlobalBase
 	}
 
@@ -184,7 +185,7 @@ func (m *TextModel) Forward(ctx ml.Context, inputs, positions, outputs ml.Tensor
 		// gemma alternates between the sliding window (local) and causal (global)
 		// kv cache every 6 layers
 		cacheType := cacheTypeSWA
-		if (i+1)%6 == 0 {
+		if (i+1)%gemmaGlobalCacheCount == 0 {
 			cacheType = cacheTypeCausal
 		}
 		cache.SetLayer(i)