Upload 6 files

chat_template.jinja

@@ -0,0 +1,75 @@
+{{- bos_token -}}
+{%- set ns = namespace(sys="", last_asst=-1) -%}
+
+{%- if messages and messages[0]["role"] == "system" -%}
+    {%- set c = messages[0]["content"] -%}
+    {%- if c is string -%}
+        {%- set ns.sys = c -%}
+    {%- else -%}
+        {%- set parts = [] -%}
+        {%- for i in c if i["type"] == "text" -%}
+            {%- set _ = parts.append(i["text"]) -%}
+        {%- endfor -%}
+        {%- set ns.sys = parts | join("") -%}
+    {%- endif -%}
+    {%- set messages = messages[1:] -%}
+{%- else -%}
+    {%- set ns.sys = "You're a helpful AI assistant." -%}
+{%- endif -%}
+
+{%- if tools -%}
+    {%- set tool_text = [] -%}
+    {%- for t in tools -%}
+        {%- set _ = tool_text.append(t if t is string else t|tojson) -%}
+    {%- endfor -%}
+    {%- set ns.sys = ns.sys + "
+
+Tools:
+" + (tool_text | join("
+")) -%}
+{%- endif -%}
+
+{{- "system
+" + ns.sys.strip() + eos_token -}}
+
+{%- for m in messages if m["role"] == "assistant" -%}
+    {%- set ns.last_asst = loop.index0 -%}
+{%- endfor -%}
+
+{%- for m in messages -%}
+    {%- set c = m["content"] -%}
+
+    {%- if c is not string -%}
+        {%- set tmp = [] -%}
+        {%- for i in c -%}
+            {%- if i["type"] == "text" -%}
+                {%- set _ = tmp.append(i["text"]) -%}
+            {%- elif i["type"] == "image" -%}
+                {%- set _ = tmp.append("<image>") -%}
+            {%- endif -%}
+        {%- endfor -%}
+        {%- set c = tmp | join("") -%}
+    {%- endif -%}
+
+    {%- if m["role"] == "assistant" and loop.index0 != ns.last_asst and "</think>" in c -%}
+        {%- set parts = c.split("</think>") -%}
+        {%- set think_lines = parts[0].split("
+") -%}
+        {%- set tail = (think_lines[-2:] | join("
+")).strip() -%}
+        {%- set c = (tail + "
+" if tail else "") + parts[-1].strip() -%}
+    {%- endif -%}
+
+    {%- set c = c.strip() -%}
+
+    {%- if c -%}
+{{- m["role"] + "
+" + c + eos_token -}}
+    {%- endif -%}
+{%- endfor -%}
+
+{%- if add_generation_prompt -%}
+{{- "assistant
+" -}}
+{%- endif -%}

config.json

@@ -0,0 +1,68 @@
+{
+  "architectures": [
+    "Gemma4ForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "attention_k_eq_v": false,
+  "bos_token_id": 1,
+  "dtype": "float32",
+  "enable_moe_block": false,
+  "eos_token_id": 2,
+  "final_logit_softcapping": null,
+  "global_head_dim": 512,
+  "head_dim": 64,
+  "hidden_activation": "gelu_pytorch_tanh",
+  "hidden_size": 896,
+  "hidden_size_per_layer_input": 8,
+  "initializer_range": 0.02,
+  "intermediate_size": 3584,
+  "layer_types": [
+    "sliding_attention",
+    "sliding_attention",
+    "sliding_attention",
+    "sliding_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "sliding_attention",
+    "sliding_attention",
+    "sliding_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention"
+  ],
+  "max_position_embeddings": 32768,
+  "model_type": "gemma4_text",
+  "moe_intermediate_size": null,
+  "norm_topk_prob": true,
+  "num_attention_heads": 14,
+  "num_experts": null,
+  "num_global_key_value_heads": null,
+  "num_hidden_layers": 14,
+  "num_key_value_heads": 2,
+  "num_kv_shared_layers": 0,
+  "pad_token_id": 0,
+  "rms_norm_eps": 1e-06,
+  "rope_parameters": {
+    "full_attention": {
+      "partial_rotary_factor": 0.25,
+      "rope_theta": 1000000.0,
+      "rope_type": "proportional"
+    },
+    "sliding_attention": {
+      "rope_theta": 10000.0,
+      "rope_type": "default"
+    }
+  },
+  "sliding_window": 128,
+  "tie_word_embeddings": true,
+  "top_k_experts": null,
+  "transformers_version": "5.8.0.dev0",
+  "use_bidirectional_attention": null,
+  "use_cache": false,
+  "use_double_wide_mlp": false,
+  "vocab_size": 24000,
+  "vocab_size_per_layer_input": 262144
+}

generation_config.json

@@ -0,0 +1,13 @@
+{
+  "_from_model_config": true,
+  "bos_token_id": 1,
+  "eos_token_id": [
+    2,
+    1
+  ],
+  "output_attentions": false,
+  "output_hidden_states": false,
+  "pad_token_id": 0,
+  "transformers_version": "5.8.0.dev0",
+  "use_cache": true
+}

model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e8552ab67404ed820154e25550f8bd54209421811160ed3da5ca8c59bd344cb8
+size 1001353760

tokenizer_config.json

@@ -0,0 +1,16 @@
+{
+  "backend": "tokenizers",
+  "bos_token": "<|sos|>",
+  "eos_token": "<|eos|>",
+  "is_local": true,
+  "local_files_only": false,
+  "max_length": 32768,
+  "model_max_length": 1000000000000000019884624838656,
+  "pad_token": "<|pad|>",
+  "padding_side": "right",
+  "stride": 0,
+  "tokenizer_class": "TokenizersBackend",
+  "truncation_side": "right",
+  "truncation_strategy": "longest_first",
+  "unk_token": "<|unk|>"
+}