Update tiny models for AXK2ForCausalLM

.gitattributes

@@ -33,3 +33,4 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
 *.zip filter=lfs diff=lfs merge=lfs -text
 *.zst filter=lfs diff=lfs merge=lfs -text
 *tfevents* filter=lfs diff=lfs merge=lfs -text
+tokenizer.json filter=lfs diff=lfs merge=lfs -text

chat_template.jinja

@@ -0,0 +1,145 @@
+{%- if tools is iterable and tools | length > 0 %}
+{{- '<|im_start|>system
+당신은 도구 호출 기능을 갖춘 유용한 도우미입니다. 사용자의 요청을 처리하기 위해서 필요한 도구가 주어진 목록에 있는 경우 도구 호출로 응답하세요.
+필요한 도구가 목록에 없는 경우에는 도구 호출 없이 사용자가 요구한 정보를 제공하세요.
+필요한 도구가 목록에 있지만 해당 도구를 호출하는데 필요한 argument 정보가 부족한 경우 해당 정보를 사용자에게 요청하세요.
+사용자의 요청을 처리하기 위해 여러번 도구를 호출할 수 있어야 합니다.
+도구 호출 이후 도구 실행 결과를 입력으로 받으면 해당 결과를 활용하여 답변을 생성하세요.
+
+다음은 접근할 수 있는 도구들의 목록 입니다:
+<tools>
+' }}
+    {%- for t in tools %}
+        {{- t | tojson }}
+        {{- '\n' }}
+    {%- endfor %}
+{{- '</tools>
+
+도구를 호출하려면 아래의 JSON으로 응답하세요.
+도구 호출 형식: <tool_call>{"name": 도구 이름, "arguments": dictionary 형태의 도구 인자값}</tool_call><|im_end|>
+' }}
+{%- endif %}
+
+{%- macro render_content(content) -%}
+    {#- `content` may be a plain string, none, or a list of parts as sent by
+        OpenAI-compatible clients: [{"type": "text", "text": "..."}] -#}
+    {%- if content is none -%}
+    {%- elif content is string -%}
+        {{- content -}}
+    {%- elif content is mapping -%}
+        {{- content.text if content.text is defined else content -}}
+    {%- elif content is sequence -%}
+        {%- for part in content -%}
+            {%- if part is mapping and part.text is defined -%}
+                {{- part.text -}}
+            {%- elif part is string -%}
+                {{- part -}}
+            {%- endif -%}
+        {%- endfor -%}
+    {%- else -%}
+        {{- content -}}
+    {%- endif -%}
+{%- endmacro -%}
+
+{%- macro render_tool_calls(tool_calls) -%}
+    {%- for tool_call in tool_calls -%}
+        {%- if tool_call.function is defined -%}
+            {%- set tool_call = tool_call.function -%}
+        {%- endif -%}
+        {{- '<tool_call>\n' -}}
+        {{- '{"name": "' + tool_call.name + '"' -}}
+        {%- if tool_call.arguments is defined -%}
+            {{- ', "arguments": ' -}}
+            {#- OpenAI-compatible clients send `arguments` as a JSON string; pass it through -#}
+            {%- if tool_call.arguments is string -%}
+                {{- tool_call.arguments -}}
+            {%- else -%}
+                {{- tool_call.arguments | tojson -}}
+            {%- endif -%}
+        {%- endif -%}
+        {{- '}\n</tool_call>' -}}
+    {%- endfor -%}
+{%- endmacro -%}
+
+{%- set ns = namespace(last_user_index=-1) %}
+{%- for m in messages %}
+    {%- if m.role == 'user' %}
+        {%- set ns.last_user_index = loop.index0 %}
+    {%- endif %}
+{%- endfor %}
+
+{%- for message in messages %}
+    {%- if message.role == 'system' or message.role == 'user' %}
+        {{- '<|im_start|>' + message.role + '\n' }}
+        {%- set reasoning_content = '' %}
+        {%- if message.reasoning_content is defined and message.reasoning_content is not none %}
+            {%- set reasoning_content = message.reasoning_content %}
+        {%- elif message.reasoning is defined and message.reasoning is not none %}
+            {%- set reasoning_content = message.reasoning %}
+        {%- endif %}
+        {%- if reasoning_content %}
+            {{- '<think>' + reasoning_content.strip() + '</think>' }}
+        {%- endif %}
+        {{- render_content(message.content) }}
+	{%- if message.tool_calls is defined and message.tool_calls is not none %}
+            {{- render_tool_calls(message.tool_calls) }}
+        {%- endif %}
+        {{- '<|im_end|>\n' }}
+    {%- elif message.role == 'assistant' %}
+        {{- '<|im_start|>assistant\n' }}
+
+        {#- Extract reasoning and content -#}
+        {%- set reasoning_content = '' %}
+        {%- set content = '' %}
+        {%- if message.content is defined and message.content is not none %}
+            {%- set content = render_content(message.content) %}
+        {%- endif %}
+        {%- if message.reasoning_content is defined and message.reasoning_content is not none %}
+            {%- set reasoning_content = message.reasoning_content %}
+        {%- elif message.reasoning is defined and message.reasoning is not none %}
+            {%- set reasoning_content = message.reasoning %}
+        {%- elif '</think>' in content %}
+            {%- set reasoning_content = content.split('</think>')[0].split('<think>')[-1] %}
+            {%- set content = content.split('</think>')[-1].lstrip('\n') %}
+        {%- endif %}
+
+        {#- Determine whether to render thinking -#}
+        {%- set render_thinking = false %}
+        {%- if reasoning_content %}
+            {%- if show_thinking is defined and show_thinking %}
+                {%- set render_thinking = true %}
+            {%- elif not (show_thinking is defined and not show_thinking) %}
+                {#- show_thinking undefined: show if last turn or tool-call turn after last user -#}
+                {%- if loop.last or (message.tool_calls is defined and loop.index0 > ns.last_user_index) %}
+                    {%- set render_thinking = true %}
+                {%- endif %}
+            {%- endif %}
+        {%- endif %}
+        {%- if render_thinking %}
+            {{- '<think>' + reasoning_content.strip() + '</think>' }}
+        {%- else %}
+            {{- '</think>' }}
+        {%- endif %}
+
+        {{- content }}
+	{%- if message.tool_calls is defined and message.tool_calls is not none %}
+            {{- render_tool_calls(message.tool_calls) }}
+        {%- endif %}
+        {{- '<|im_end|>' + ('\n' if (not loop.last or add_generation_prompt) else '') }}
+    {%- elif message.role == 'tool' %}
+        {{- '<|im_start|>tool\n' }}
+        {{- '<tool_response>\n' }}
+        {{- render_content(message.content) }}
+        {{- '\n' }}
+        {{- '</tool_response><|im_end|>\n' }}
+    {%- endif %}
+{%- endfor %}
+
+{%- if add_generation_prompt %}
+    {{- '<|im_start|>assistant\n' }}
+    {%- if enable_thinking is defined and enable_thinking is true %}
+        {{- '<think>' }}
+    {%- else %}
+        {{- '</think>' }}
+    {%- endif %}
+{%- endif %}

config.json

@@ -0,0 +1,57 @@
+{
+  "architectures": [
+    "AXK2ForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "bos_token_id": 163691,
+  "dtype": "float32",
+  "eos_token_id": 163704,
+  "gated_norm_rank": 4,
+  "head_dim": 64,
+  "hidden_act": "gelu",
+  "hidden_size": 32,
+  "index_head_dim": 64,
+  "index_n_heads": 2,
+  "index_topk": 8,
+  "initializer_range": 0.02,
+  "intermediate_size": 32,
+  "is_decoder": true,
+  "kv_lora_rank": 32,
+  "layer_types": [
+    "deepseek_sparse_attention",
+    "deepseek_sparse_attention"
+  ],
+  "max_position_embeddings": 512,
+  "mlp_layer_types": [
+    "dense",
+    "sparse"
+  ],
+  "model_type": "axk2",
+  "moe_intermediate_size": 16,
+  "n_group": null,
+  "n_routed_experts": 8,
+  "n_shared_experts": 1,
+  "norm_topk_prob": true,
+  "num_attention_heads": 2,
+  "num_experts_per_tok": 2,
+  "num_hidden_layers": 2,
+  "num_key_value_heads": 2,
+  "pad_token_id": 163692,
+  "q_lora_rank": 16,
+  "qk_head_dim": 128,
+  "qk_nope_head_dim": 64,
+  "qk_rope_head_dim": 64,
+  "rms_norm_eps": 1e-06,
+  "rope_parameters": {
+    "rope_theta": 10000.0,
+    "rope_type": "default"
+  },
+  "routed_scaling_factor": 2.5,
+  "tie_word_embeddings": false,
+  "topk_group": null,
+  "transformers_version": "5.16.0.dev0",
+  "use_cache": true,
+  "v_head_dim": 32,
+  "vocab_size": 163840
+}

generation_config.json

@@ -0,0 +1,10 @@
+{
+  "_from_model_config": true,
+  "bos_token_id": 163691,
+  "eos_token_id": 163704,
+  "output_attentions": false,
+  "output_hidden_states": false,
+  "pad_token_id": 163692,
+  "transformers_version": "5.16.0.dev0",
+  "use_cache": true
+}

model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d8bb6e9119100c89e61ffe03edc0ba789f479ae635bfbdf83871d1034cc6fe75
+size 42234000

tokenizer.json

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:af7a63a978968e122296257f015acaf6412a0c57832ead98de556bf740f1750b
+size 13210021

tokenizer_config.json

@@ -0,0 +1,21 @@
+{
+  "backend": "tokenizers",
+  "bos_token": "<|endoftext|>",
+  "clean_up_tokenization_spaces": true,
+  "cls_token": "<|cls|>",
+  "eod_token": "<|endoftext|>",
+  "eos_token": "<|im_end|>",
+  "errors": "replace",
+  "is_local": true,
+  "local_files_only": false,
+  "mask_token": "<|mask|>",
+  "model_max_length": 512,
+  "model_specific_special_tokens": {
+    "eod_token": "<|endoftext|>"
+  },
+  "pad_token": "<|pad|>",
+  "sep_token": "<|sep|>",
+  "tokenizer_class": "TokenizersBackend",
+  "unk_token": "<|unk|>",
+  "vocab_size": 163840
+}