Upload folder using huggingface_hub

.gitattributes

@@ -33,3 +33,7 @@ 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
+checkpoint-200/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+checkpoint-400/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+checkpoint-600/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+tokenizer.json filter=lfs diff=lfs merge=lfs -text

README.md

@@ -0,0 +1,63 @@
+---
+base_model: chenhaodev/unsloth_gemma-4-E4B-medical-o1-reasoning-SFT-merged
+library_name: peft
+model_name: chenhaodev_unsloth_gemma-4-E4B-medical-o1-reasoning-SFT-merged_1776696734
+tags:
+- base_model:adapter:chenhaodev/unsloth_gemma-4-E4B-medical-o1-reasoning-SFT-merged
+- lora
+- sft
+- transformers
+- trl
+- unsloth
+licence: license
+pipeline_tag: text-generation
+---
+
+# Model Card for chenhaodev_unsloth_gemma-4-E4B-medical-o1-reasoning-SFT-merged_1776696734
+
+This model is a fine-tuned version of [chenhaodev/unsloth_gemma-4-E4B-medical-o1-reasoning-SFT-merged](https://huggingface.co/chenhaodev/unsloth_gemma-4-E4B-medical-o1-reasoning-SFT-merged).
+It has been trained using [TRL](https://github.com/huggingface/trl).
+
+## Quick start
+
+```python
+from transformers import pipeline
+
+question = "If you had a time machine, but could only go to the past or the future once and never return, which would you choose and why?"
+generator = pipeline("text-generation", model="None", device="cuda")
+output = generator([{"role": "user", "content": question}], max_new_tokens=128, return_full_text=False)[0]
+print(output["generated_text"])
+```
+
+## Training procedure
+
+ 
+
+
+This model was trained with SFT.
+
+### Framework versions
+
+- PEFT 0.18.1
+- TRL: 0.23.1
+- Transformers: 5.5.0
+- Pytorch: 2.10.0+cu128
+- Datasets: 4.3.0
+- Tokenizers: 0.22.2
+
+## Citations
+
+
+
+Cite TRL as:
+    
+```bibtex
+@misc{vonwerra2022trl,
+	title        = {{TRL: Transformer Reinforcement Learning}},
+	author       = {Leandro von Werra and Younes Belkada and Lewis Tunstall and Edward Beeching and Tristan Thrush and Nathan Lambert and Shengyi Huang and Kashif Rasul and Quentin Gallou{\'e}dec},
+	year         = 2020,
+	journal      = {GitHub repository},
+	publisher    = {GitHub},
+	howpublished = {\url{https://github.com/huggingface/trl}}
+}
+```

adapter_config.json

@@ -0,0 +1,51 @@
+{
+  "alora_invocation_tokens": null,
+  "alpha_pattern": {},
+  "arrow_config": null,
+  "auto_mapping": {
+    "base_model_class": "Gemma4ForConditionalGeneration",
+    "parent_library": "transformers.models.gemma4.modeling_gemma4",
+    "unsloth_fixed": true
+  },
+  "base_model_name_or_path": "chenhaodev/unsloth_gemma-4-E4B-medical-o1-reasoning-SFT-merged",
+  "bias": "none",
+  "corda_config": null,
+  "ensure_weight_tying": false,
+  "eva_config": null,
+  "exclude_modules": null,
+  "fan_in_fan_out": false,
+  "inference_mode": true,
+  "init_lora_weights": true,
+  "layer_replication": null,
+  "layers_pattern": null,
+  "layers_to_transform": null,
+  "loftq_config": {},
+  "lora_alpha": 32,
+  "lora_bias": false,
+  "lora_dropout": 0.0,
+  "megatron_config": null,
+  "megatron_core": "megatron.core",
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "peft_version": "0.18.1",
+  "qalora_group_size": 16,
+  "r": 16,
+  "rank_pattern": {},
+  "revision": null,
+  "target_modules": [
+    "o_proj",
+    "down_proj",
+    "k_proj",
+    "up_proj",
+    "q_proj",
+    "v_proj",
+    "gate_proj"
+  ],
+  "target_parameters": null,
+  "task_type": "CAUSAL_LM",
+  "trainable_token_indices": null,
+  "use_dora": false,
+  "use_qalora": false,
+  "use_rslora": false,
+  "unsloth_training_method": "qlora"
+}

adapter_model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:74cbeea179026dd191ebed8036ca25bfe0e8a34550ca98a20fb31cecc76780c4
+size 169741912

chat_template.jinja

@@ -0,0 +1,344 @@
+{%- macro format_parameters(properties, required) -%}
+    {%- set standard_keys = ['description', 'type', 'properties', 'required', 'nullable'] -%}
+    {%- set ns = namespace(found_first=false) -%}
+    {%- for key, value in properties | dictsort -%}
+        {%- set add_comma = false -%}
+        {%- if key not in standard_keys -%}
+            {%- if ns.found_first %},{% endif -%}
+            {%- set ns.found_first = true -%}
+            {{ key }}:{
+            {%- if value['description'] -%}
+                description:<|"|>{{ value['description'] }}<|"|>
+                {%- set add_comma = true -%}
+            {%- endif -%}
+            {%- if value['type'] | upper == 'STRING' -%}
+                {%- if value['enum'] -%}
+                    {%- if add_comma %},{%- else -%} {%- set add_comma = true -%} {% endif -%}
+                    enum:{{ format_argument(value['enum']) }}
+                {%- endif -%}
+            {%- elif value['type'] | upper == 'ARRAY' -%}
+                {%- if value['items'] is mapping and value['items'] -%}
+                    {%- if add_comma %},{%- else -%} {%- set add_comma = true -%} {% endif -%}
+                    items:{
+                    {%- set ns_items = namespace(found_first=false) -%}
+                    {%- for item_key, item_value in value['items'] | dictsort -%}
+                        {%- if item_value is not none -%}
+                            {%- if ns_items.found_first %},{% endif -%}
+                            {%- set ns_items.found_first = true -%}
+                            {%- if item_key == 'properties' -%}
+                                properties:{
+                                {%- if item_value is mapping -%}
+                                    {{- format_parameters(item_value, value['items']['required'] | default([])) -}}
+                                {%- endif -%}
+                                }
+                            {%- elif item_key == 'required' -%}
+                                required:[
+                                {%- for req_item in item_value -%}
+                                    <|"|>{{- req_item -}}<|"|>
+                                    {%- if not loop.last %},{% endif -%}
+                                {%- endfor -%}
+                                ]
+                            {%- elif item_key == 'type' -%}
+                                {%- if item_value is string -%}
+                                    type:{{ format_argument(item_value | upper) }}
+                                {%- else -%}
+                                    type:{{ format_argument(item_value | map('upper') | list) }}
+                                {%- endif -%}
+                            {%- else -%}
+                                {{ item_key }}:{{ format_argument(item_value) }}
+                            {%- endif -%}
+                        {%- endif -%}
+                    {%- endfor -%}
+                    }
+                {%- endif -%}
+            {%- endif -%}
+            {%- if value['nullable'] %}
+                {%- if add_comma %},{%- else -%} {%- set add_comma = true -%} {% endif -%}
+                nullable:true
+            {%- endif -%}
+            {%- if value['type'] | upper == 'OBJECT' -%}
+                {%- if value['properties'] is defined and value['properties'] is mapping -%}
+                    {%- if add_comma %},{%- else -%} {%- set add_comma = true -%} {% endif -%}
+                    properties:{
+                    {{- format_parameters(value['properties'], value['required'] | default([])) -}}
+                    }
+                {%- elif value is mapping -%}
+                    {%- if add_comma %},{%- else -%} {%- set add_comma = true -%} {% endif -%}
+                    properties:{
+                    {{- format_parameters(value, value['required'] | default([])) -}}
+                    }
+                {%- endif -%}
+                {%- if value['required'] -%}
+                    {%- if add_comma %},{%- else -%} {%- set add_comma = true -%} {% endif -%}
+                    required:[
+                    {%- for item in value['required'] | default([]) -%}
+                        <|"|>{{- item -}}<|"|>
+                        {%- if not loop.last %},{% endif -%}
+                    {%- endfor -%}
+                    ]
+                {%- endif -%}
+            {%- endif -%}
+            {%- if add_comma %},{%- else -%} {%- set add_comma = true -%} {% endif -%}
+            type:<|"|>{{ value['type'] | upper }}<|"|>}
+        {%- endif -%}
+    {%- endfor -%}
+{%- endmacro -%}
+{%- macro format_function_declaration(tool_data) -%}
+    declaration:{{- tool_data['function']['name'] -}}{description:<|"|>{{- tool_data['function']['description'] -}}<|"|>
+    {%- set params = tool_data['function']['parameters'] -%}
+    {%- if params -%}
+        ,parameters:{
+        {%- if params['properties'] -%}
+            properties:{ {{- format_parameters(params['properties'], params['required']) -}} },
+        {%- endif -%}
+        {%- if params['required'] -%}
+            required:[
+            {%- for item in params['required'] -%}
+                <|"|>{{- item -}}<|"|>
+                {{- ',' if not loop.last -}}
+            {%- endfor -%}
+            ],
+        {%- endif -%}
+        {%- if params['type'] -%}
+            type:<|"|>{{- params['type'] | upper -}}<|"|>}
+        {%- endif -%}
+    {%- endif -%}
+    {%- if 'response' in tool_data['function'] -%}
+        {%- set response_declaration = tool_data['function']['response'] -%}
+        ,response:{
+        {%- if response_declaration['description'] -%}
+            description:<|"|>{{- response_declaration['description'] -}}<|"|>,
+        {%- endif -%}
+        {%- if response_declaration['type'] | upper == 'OBJECT' -%}
+            type:<|"|>{{- response_declaration['type'] | upper -}}<|"|>}
+        {%- endif -%}
+    {%- endif -%}
+    }
+{%- endmacro -%}
+{%- macro format_argument(argument, escape_keys=True) -%}
+    {%- if argument is string -%}
+        {{- '<|"|>' + argument + '<|"|>' -}}
+    {%- elif argument is boolean -%}
+        {{- 'true' if argument else 'false' -}}
+    {%- elif argument is mapping -%}
+        {{- '{' -}}
+        {%- set ns = namespace(found_first=false) -%}
+        {%- for key, value in argument | dictsort -%}
+            {%- if ns.found_first %},{% endif -%}
+            {%- set ns.found_first = true -%}
+            {%- if escape_keys -%}
+                {{- '<|"|>' + key + '<|"|>' -}}
+            {%- else -%}
+                {{- key -}}
+            {%- endif -%}
+            :{{- format_argument(value, escape_keys=escape_keys) -}}
+        {%- endfor -%}
+        {{- '}' -}}
+    {%- elif argument is sequence -%}
+        {{- '[' -}}
+        {%- for item in argument -%}
+            {{- format_argument(item, escape_keys=escape_keys) -}}
+            {%- if not loop.last %},{% endif -%}
+        {%- endfor -%}
+        {{- ']' -}}
+    {%- else -%}
+        {{- argument -}}
+    {%- endif -%}
+{%- endmacro -%}
+{%- macro strip_thinking(text) -%}
+    {%- set ns = namespace(result='') -%}
+    {%- for part in text.split('<channel|>') -%}
+        {%- if '<|channel>' in part -%}
+            {%- set ns.result = ns.result + part.split('<|channel>')[0] -%}
+        {%- else -%}
+            {%- set ns.result = ns.result + part -%}
+        {%- endif -%}
+    {%- endfor -%}
+    {{- ns.result | trim -}}
+{%- endmacro -%}
+
+{%- macro format_tool_response_block(tool_name, response) -%}
+    {{- '<|tool_response>' -}}
+    {%- if response is mapping -%}
+        {{- 'response:' + tool_name + '{' -}}
+        {%- for key, value in response | dictsort -%}
+            {{- key -}}:{{- format_argument(value, escape_keys=False) -}}
+            {%- if not loop.last %},{% endif -%}
+        {%- endfor -%}
+        {{- '}' -}}
+    {%- else -%}
+        {{- 'response:' + tool_name + '{value:' + format_argument(response, escape_keys=False) + '}' -}}
+    {%- endif -%}
+    {{- '<tool_response|>' -}}
+{%- endmacro -%}
+
+{%- set ns = namespace(prev_message_type=None) -%}
+{%- set loop_messages = messages -%}
+{{- bos_token -}}
+{#- Handle System/Tool Definitions Block -#}
+{%- if (enable_thinking is defined and enable_thinking) or tools or messages[0]['role'] in ['system', 'developer'] -%}
+    {{- '<|turn>system\n' -}}
+
+    {#- Inject Thinking token at the very top of the FIRST system turn -#}
+    {%- if enable_thinking is defined and enable_thinking -%}
+        {{- '<|think|>\n' -}}
+        {%- set ns.prev_message_type = 'think' -%}
+    {%- endif -%}
+
+    {%- if messages[0]['role'] in ['system', 'developer'] -%}
+        {{- messages[0]['content'] | trim -}}
+        {%- set loop_messages = messages[1:] -%}
+    {%- endif -%}
+
+    {%- if tools -%}
+        {%- for tool in tools %}
+            {{- '<|tool>' -}}
+            {{- format_function_declaration(tool) | trim -}}
+            {{- '<tool|>' -}}
+        {%- endfor %}
+        {%- set ns.prev_message_type = 'tool' -%}
+    {%- endif -%}
+
+    {{- '<turn|>\n' -}}
+{%- endif %}
+
+{#- Pre-scan: find last user message index for reasoning guard -#}
+{%- set ns_turn = namespace(last_user_idx=-1) -%}
+{%- for i in range(loop_messages | length) -%}
+    {%- if loop_messages[i]['role'] == 'user' -%}
+        {%- set ns_turn.last_user_idx = i -%}
+    {%- endif -%}
+{%- endfor -%}
+
+{#- Loop through messages -#}
+{%- for message in loop_messages -%}
+    {%- if message['role'] != 'tool' -%}
+    {%- set ns.prev_message_type = None -%}
+    {%- set role = 'model' if message['role'] == 'assistant' else message['role'] -%}
+    {#- Detect continuation: suppress duplicate <|turn>model when previous non-tool message was also assistant -#}
+    {%- set prev_nt = namespace(role=None, found=false) -%}
+    {%- if loop.index0 > 0 -%}
+        {%- for j in range(loop.index0 - 1, -1, -1) -%}
+            {%- if not prev_nt.found -%}
+                {%- if loop_messages[j]['role'] != 'tool' -%}
+                    {%- set prev_nt.role = loop_messages[j]['role'] -%}
+                    {%- set prev_nt.found = true -%}
+                {%- endif -%}
+            {%- endif -%}
+        {%- endfor -%}
+    {%- endif -%}
+    {%- set continue_same_model_turn = (role == 'model' and prev_nt.role == 'assistant') -%}
+    {%- if not continue_same_model_turn -%}
+        {{- '<|turn>' + role + '\n' }}
+    {%- endif -%}
+
+    {#- Render reasoning/reasoning_content as thinking channel -#}
+    {%- set thinking_text = message.get('reasoning') or message.get('reasoning_content') -%}
+    {%- if thinking_text and loop.index0 > ns_turn.last_user_idx and message.get('tool_calls') -%}
+        {{- '<|channel>thought\n' + thinking_text + '\n<channel|>' -}}
+    {%- endif -%}
+
+            {%- if message['tool_calls'] -%}
+                {%- for tool_call in message['tool_calls'] -%}
+                    {%- set function = tool_call['function'] -%}
+                    {{- '<|tool_call>call:' + function['name'] + '{' -}}
+                    {%- if function['arguments'] is mapping -%}
+                        {%- set ns_args = namespace(found_first=false) -%}
+                        {%- for key, value in function['arguments'] | dictsort -%}
+                            {%- if ns_args.found_first %},{% endif -%}
+                            {%- set ns_args.found_first = true -%}
+                            {{- key -}}:{{- format_argument(value, escape_keys=False) -}}
+                        {%- endfor -%}
+                    {%- elif function['arguments'] is string -%}
+                        {{- function['arguments'] -}}
+                    {%- endif -%}
+                    {{- '}<tool_call|>' -}}
+                {%- endfor -%}
+                {%- set ns.prev_message_type = 'tool_call' -%}
+            {%- endif -%}
+
+            {%- set ns_tr_out = namespace(flag=false) -%}
+            {%- if message.get('tool_responses') -%}
+                {#- Legacy: tool_responses embedded on the assistant message (Google/Gemma native) -#}
+                {%- for tool_response in message['tool_responses'] -%}
+                    {{- format_tool_response_block(tool_response['name'] | default('unknown'), tool_response['response']) -}}
+                    {%- set ns_tr_out.flag = true -%}
+                    {%- set ns.prev_message_type = 'tool_response' -%}
+                {%- endfor -%}
+            {%- elif message.get('tool_calls') -%}
+                {#- OpenAI Chat Completions: forward-scan consecutive role:tool messages -#}
+                {%- set ns_tool_scan = namespace(stopped=false) -%}
+                {%- for k in range(loop.index0 + 1, loop_messages | length) -%}
+                    {%- if ns_tool_scan.stopped -%}
+                    {%- elif loop_messages[k]['role'] != 'tool' -%}
+                        {%- set ns_tool_scan.stopped = true -%}
+                    {%- else -%}
+                        {%- set follow = loop_messages[k] -%}
+                        {#- Resolve tool_call_id to function name -#}
+                        {%- set ns_tname = namespace(name=follow.get('name') | default('unknown')) -%}
+                        {%- for tc in message['tool_calls'] -%}
+                            {%- if tc.get('id') == follow.get('tool_call_id') -%}
+                                {%- set ns_tname.name = tc['function']['name'] -%}
+                            {%- endif -%}
+                        {%- endfor -%}
+                        {#- Handle content as string or content-parts array -#}
+                        {%- set tool_body = follow.get('content') -%}
+                        {%- if tool_body is string -%}
+                            {{- format_tool_response_block(ns_tname.name, tool_body) -}}
+                        {%- elif tool_body is sequence and tool_body is not string -%}
+                            {%- set ns_txt = namespace(s='') -%}
+                            {%- for part in tool_body -%}
+                                {%- if part.get('type') == 'text' -%}
+                                    {%- set ns_txt.s = ns_txt.s + (part.get('text') | default('')) -%}
+                                {%- endif -%}
+                            {%- endfor -%}
+                            {{- format_tool_response_block(ns_tname.name, ns_txt.s) -}}
+                        {%- else -%}
+                            {{- format_tool_response_block(ns_tname.name, tool_body) -}}
+                        {%- endif -%}
+                        {%- set ns_tr_out.flag = true -%}
+                        {%- set ns.prev_message_type = 'tool_response' -%}
+                    {%- endif -%}
+                {%- endfor -%}
+            {%- endif -%}
+
+            {%- if message['content'] is string -%}
+                {%- if role == 'model' -%}
+                    {{- strip_thinking(message['content']) -}}
+                {%- else -%}
+                    {{- message['content'] | trim -}}
+                {%- endif -%}
+            {%- elif message['content'] is sequence -%}
+                {%- for item in message['content'] -%}
+                    {%- if item['type'] == 'text' -%}
+                        {%- if role == 'model' -%}
+                            {{- strip_thinking(item['text']) -}}
+                        {%- else -%}
+                            {{- item['text'] | trim -}}
+                        {%- endif -%}
+                    {%- elif item['type'] == 'image' -%}
+                        {{- '<|image|>' -}}
+                        {%- set ns.prev_message_type = 'image' -%}
+                    {%- elif item['type'] == 'audio' -%}
+                        {{- '<|audio|>' -}}
+                        {%- set ns.prev_message_type = 'audio' -%}
+                    {%- elif item['type'] == 'video' -%}
+                        {{- '<|video|>' -}}
+                        {%- set ns.prev_message_type = 'video' -%}
+                    {%- endif -%}
+                {%- endfor -%}
+            {%- endif -%}
+
+        {%- if ns.prev_message_type == 'tool_call' and not ns_tr_out.flag -%}
+            {{- '<|tool_response>' -}}
+        {%- elif not (ns_tr_out.flag and not message.get('content')) -%}
+            {{- '<turn|>\n' -}}
+        {%- endif -%}
+    {%- endif -%}
+{%- endfor -%}
+
+{%- if add_generation_prompt -%}
+    {%- if ns.prev_message_type != 'tool_response' and ns.prev_message_type != 'tool_call' -%}
+        {{- '<|turn>model\n' -}}
+    {%- endif -%}
+{%- endif -%}

checkpoint-200/README.md

@@ -0,0 +1,210 @@
+---
+base_model: chenhaodev/unsloth_gemma-4-E4B-medical-o1-reasoning-SFT-merged
+library_name: peft
+pipeline_tag: text-generation
+tags:
+- base_model:adapter:chenhaodev/unsloth_gemma-4-E4B-medical-o1-reasoning-SFT-merged
+- lora
+- sft
+- transformers
+- trl
+- unsloth
+---
+
+# Model Card for Model ID
+
+<!-- Provide a quick summary of what the model is/does. -->
+
+
+
+## Model Details
+
+### Model Description
+
+<!-- Provide a longer summary of what this model is. -->
+
+
+
+- **Developed by:** [More Information Needed]
+- **Funded by [optional]:** [More Information Needed]
+- **Shared by [optional]:** [More Information Needed]
+- **Model type:** [More Information Needed]
+- **Language(s) (NLP):** [More Information Needed]
+- **License:** [More Information Needed]
+- **Finetuned from model [optional]:** [More Information Needed]
+
+### Model Sources [optional]
+
+<!-- Provide the basic links for the model. -->
+
+- **Repository:** [More Information Needed]
+- **Paper [optional]:** [More Information Needed]
+- **Demo [optional]:** [More Information Needed]
+
+## Uses
+
+<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
+
+### Direct Use
+
+<!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
+
+[More Information Needed]
+
+### Downstream Use [optional]
+
+<!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->
+
+[More Information Needed]
+
+### Out-of-Scope Use
+
+<!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
+
+[More Information Needed]
+
+## Bias, Risks, and Limitations
+
+<!-- This section is meant to convey both technical and sociotechnical limitations. -->
+
+[More Information Needed]
+
+### Recommendations
+
+<!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->
+
+Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
+
+## How to Get Started with the Model
+
+Use the code below to get started with the model.
+
+[More Information Needed]
+
+## Training Details
+
+### Training Data
+
+<!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
+
+[More Information Needed]
+
+### Training Procedure
+
+<!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
+
+#### Preprocessing [optional]
+
+[More Information Needed]
+
+
+#### Training Hyperparameters
+
+- **Training regime:** [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->
+
+#### Speeds, Sizes, Times [optional]
+
+<!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
+
+[More Information Needed]
+
+## Evaluation
+
+<!-- This section describes the evaluation protocols and provides the results. -->
+
+### Testing Data, Factors & Metrics
+
+#### Testing Data
+
+<!-- This should link to a Dataset Card if possible. -->
+
+[More Information Needed]
+
+#### Factors
+
+<!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->
+
+[More Information Needed]
+
+#### Metrics
+
+<!-- These are the evaluation metrics being used, ideally with a description of why. -->
+
+[More Information Needed]
+
+### Results
+
+[More Information Needed]
+
+#### Summary
+
+
+
+## Model Examination [optional]
+
+<!-- Relevant interpretability work for the model goes here -->
+
+[More Information Needed]
+
+## Environmental Impact
+
+<!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly -->
+
+Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700).
+
+- **Hardware Type:** [More Information Needed]
+- **Hours used:** [More Information Needed]
+- **Cloud Provider:** [More Information Needed]
+- **Compute Region:** [More Information Needed]
+- **Carbon Emitted:** [More Information Needed]
+
+## Technical Specifications [optional]
+
+### Model Architecture and Objective
+
+[More Information Needed]
+
+### Compute Infrastructure
+
+[More Information Needed]
+
+#### Hardware
+
+[More Information Needed]
+
+#### Software
+
+[More Information Needed]
+
+## Citation [optional]
+
+<!-- If there is a paper or blog post introducing the model, the APA and Bibtex information for that should go in this section. -->
+
+**BibTeX:**
+
+[More Information Needed]
+
+**APA:**
+
+[More Information Needed]
+
+## Glossary [optional]
+
+<!-- If relevant, include terms and calculations in this section that can help readers understand the model or model card. -->
+
+[More Information Needed]
+
+## More Information [optional]
+
+[More Information Needed]
+
+## Model Card Authors [optional]
+
+[More Information Needed]
+
+## Model Card Contact
+
+[More Information Needed]
+### Framework versions
+
+- PEFT 0.18.1

checkpoint-200/adapter_config.json

@@ -0,0 +1,50 @@
+{
+  "alora_invocation_tokens": null,
+  "alpha_pattern": {},
+  "arrow_config": null,
+  "auto_mapping": {
+    "base_model_class": "Gemma4ForConditionalGeneration",
+    "parent_library": "transformers.models.gemma4.modeling_gemma4",
+    "unsloth_fixed": true
+  },
+  "base_model_name_or_path": "chenhaodev/unsloth_gemma-4-E4B-medical-o1-reasoning-SFT-merged",
+  "bias": "none",
+  "corda_config": null,
+  "ensure_weight_tying": false,
+  "eva_config": null,
+  "exclude_modules": null,
+  "fan_in_fan_out": false,
+  "inference_mode": true,
+  "init_lora_weights": true,
+  "layer_replication": null,
+  "layers_pattern": null,
+  "layers_to_transform": null,
+  "loftq_config": {},
+  "lora_alpha": 32,
+  "lora_bias": false,
+  "lora_dropout": 0.0,
+  "megatron_config": null,
+  "megatron_core": "megatron.core",
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "peft_version": "0.18.1",
+  "qalora_group_size": 16,
+  "r": 16,
+  "rank_pattern": {},
+  "revision": null,
+  "target_modules": [
+    "o_proj",
+    "down_proj",
+    "k_proj",
+    "up_proj",
+    "q_proj",
+    "v_proj",
+    "gate_proj"
+  ],
+  "target_parameters": null,
+  "task_type": "CAUSAL_LM",
+  "trainable_token_indices": null,
+  "use_dora": false,
+  "use_qalora": false,
+  "use_rslora": false
+}

checkpoint-200/adapter_model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3878aeff14c39e9b2d0de794d0963a43f82a41aacb8d5ebbecf5e1a4e2986ae3
+size 169741912

checkpoint-200/chat_template.jinja

@@ -0,0 +1,344 @@
+{%- macro format_parameters(properties, required) -%}
+    {%- set standard_keys = ['description', 'type', 'properties', 'required', 'nullable'] -%}
+    {%- set ns = namespace(found_first=false) -%}
+    {%- for key, value in properties | dictsort -%}
+        {%- set add_comma = false -%}
+        {%- if key not in standard_keys -%}
+            {%- if ns.found_first %},{% endif -%}
+            {%- set ns.found_first = true -%}
+            {{ key }}:{
+            {%- if value['description'] -%}
+                description:<|"|>{{ value['description'] }}<|"|>
+                {%- set add_comma = true -%}
+            {%- endif -%}
+            {%- if value['type'] | upper == 'STRING' -%}
+                {%- if value['enum'] -%}
+                    {%- if add_comma %},{%- else -%} {%- set add_comma = true -%} {% endif -%}
+                    enum:{{ format_argument(value['enum']) }}
+                {%- endif -%}
+            {%- elif value['type'] | upper == 'ARRAY' -%}
+                {%- if value['items'] is mapping and value['items'] -%}
+                    {%- if add_comma %},{%- else -%} {%- set add_comma = true -%} {% endif -%}
+                    items:{
+                    {%- set ns_items = namespace(found_first=false) -%}
+                    {%- for item_key, item_value in value['items'] | dictsort -%}
+                        {%- if item_value is not none -%}
+                            {%- if ns_items.found_first %},{% endif -%}
+                            {%- set ns_items.found_first = true -%}
+                            {%- if item_key == 'properties' -%}
+                                properties:{
+                                {%- if item_value is mapping -%}
+                                    {{- format_parameters(item_value, value['items']['required'] | default([])) -}}
+                                {%- endif -%}
+                                }
+                            {%- elif item_key == 'required' -%}
+                                required:[
+                                {%- for req_item in item_value -%}
+                                    <|"|>{{- req_item -}}<|"|>
+                                    {%- if not loop.last %},{% endif -%}
+                                {%- endfor -%}
+                                ]
+                            {%- elif item_key == 'type' -%}
+                                {%- if item_value is string -%}
+                                    type:{{ format_argument(item_value | upper) }}
+                                {%- else -%}
+                                    type:{{ format_argument(item_value | map('upper') | list) }}
+                                {%- endif -%}
+                            {%- else -%}
+                                {{ item_key }}:{{ format_argument(item_value) }}
+                            {%- endif -%}
+                        {%- endif -%}
+                    {%- endfor -%}
+                    }
+                {%- endif -%}
+            {%- endif -%}
+            {%- if value['nullable'] %}
+                {%- if add_comma %},{%- else -%} {%- set add_comma = true -%} {% endif -%}
+                nullable:true
+            {%- endif -%}
+            {%- if value['type'] | upper == 'OBJECT' -%}
+                {%- if value['properties'] is defined and value['properties'] is mapping -%}
+                    {%- if add_comma %},{%- else -%} {%- set add_comma = true -%} {% endif -%}
+                    properties:{
+                    {{- format_parameters(value['properties'], value['required'] | default([])) -}}
+                    }
+                {%- elif value is mapping -%}
+                    {%- if add_comma %},{%- else -%} {%- set add_comma = true -%} {% endif -%}
+                    properties:{
+                    {{- format_parameters(value, value['required'] | default([])) -}}
+                    }
+                {%- endif -%}
+                {%- if value['required'] -%}
+                    {%- if add_comma %},{%- else -%} {%- set add_comma = true -%} {% endif -%}
+                    required:[
+                    {%- for item in value['required'] | default([]) -%}
+                        <|"|>{{- item -}}<|"|>
+                        {%- if not loop.last %},{% endif -%}
+                    {%- endfor -%}
+                    ]
+                {%- endif -%}
+            {%- endif -%}
+            {%- if add_comma %},{%- else -%} {%- set add_comma = true -%} {% endif -%}
+            type:<|"|>{{ value['type'] | upper }}<|"|>}
+        {%- endif -%}
+    {%- endfor -%}
+{%- endmacro -%}
+{%- macro format_function_declaration(tool_data) -%}
+    declaration:{{- tool_data['function']['name'] -}}{description:<|"|>{{- tool_data['function']['description'] -}}<|"|>
+    {%- set params = tool_data['function']['parameters'] -%}
+    {%- if params -%}
+        ,parameters:{
+        {%- if params['properties'] -%}
+            properties:{ {{- format_parameters(params['properties'], params['required']) -}} },
+        {%- endif -%}
+        {%- if params['required'] -%}
+            required:[
+            {%- for item in params['required'] -%}
+                <|"|>{{- item -}}<|"|>
+                {{- ',' if not loop.last -}}
+            {%- endfor -%}
+            ],
+        {%- endif -%}
+        {%- if params['type'] -%}
+            type:<|"|>{{- params['type'] | upper -}}<|"|>}
+        {%- endif -%}
+    {%- endif -%}
+    {%- if 'response' in tool_data['function'] -%}
+        {%- set response_declaration = tool_data['function']['response'] -%}
+        ,response:{
+        {%- if response_declaration['description'] -%}
+            description:<|"|>{{- response_declaration['description'] -}}<|"|>,
+        {%- endif -%}
+        {%- if response_declaration['type'] | upper == 'OBJECT' -%}
+            type:<|"|>{{- response_declaration['type'] | upper -}}<|"|>}
+        {%- endif -%}
+    {%- endif -%}
+    }
+{%- endmacro -%}
+{%- macro format_argument(argument, escape_keys=True) -%}
+    {%- if argument is string -%}
+        {{- '<|"|>' + argument + '<|"|>' -}}
+    {%- elif argument is boolean -%}
+        {{- 'true' if argument else 'false' -}}
+    {%- elif argument is mapping -%}
+        {{- '{' -}}
+        {%- set ns = namespace(found_first=false) -%}
+        {%- for key, value in argument | dictsort -%}
+            {%- if ns.found_first %},{% endif -%}
+            {%- set ns.found_first = true -%}
+            {%- if escape_keys -%}
+                {{- '<|"|>' + key + '<|"|>' -}}
+            {%- else -%}
+                {{- key -}}
+            {%- endif -%}
+            :{{- format_argument(value, escape_keys=escape_keys) -}}
+        {%- endfor -%}
+        {{- '}' -}}
+    {%- elif argument is sequence -%}
+        {{- '[' -}}
+        {%- for item in argument -%}
+            {{- format_argument(item, escape_keys=escape_keys) -}}
+            {%- if not loop.last %},{% endif -%}
+        {%- endfor -%}
+        {{- ']' -}}
+    {%- else -%}
+        {{- argument -}}
+    {%- endif -%}
+{%- endmacro -%}
+{%- macro strip_thinking(text) -%}
+    {%- set ns = namespace(result='') -%}
+    {%- for part in text.split('<channel|>') -%}
+        {%- if '<|channel>' in part -%}
+            {%- set ns.result = ns.result + part.split('<|channel>')[0] -%}
+        {%- else -%}
+            {%- set ns.result = ns.result + part -%}
+        {%- endif -%}
+    {%- endfor -%}
+    {{- ns.result | trim -}}
+{%- endmacro -%}
+
+{%- macro format_tool_response_block(tool_name, response) -%}
+    {{- '<|tool_response>' -}}
+    {%- if response is mapping -%}
+        {{- 'response:' + tool_name + '{' -}}
+        {%- for key, value in response | dictsort -%}
+            {{- key -}}:{{- format_argument(value, escape_keys=False) -}}
+            {%- if not loop.last %},{% endif -%}
+        {%- endfor -%}
+        {{- '}' -}}
+    {%- else -%}
+        {{- 'response:' + tool_name + '{value:' + format_argument(response, escape_keys=False) + '}' -}}
+    {%- endif -%}
+    {{- '<tool_response|>' -}}
+{%- endmacro -%}
+
+{%- set ns = namespace(prev_message_type=None) -%}
+{%- set loop_messages = messages -%}
+{{- bos_token -}}
+{#- Handle System/Tool Definitions Block -#}
+{%- if (enable_thinking is defined and enable_thinking) or tools or messages[0]['role'] in ['system', 'developer'] -%}
+    {{- '<|turn>system\n' -}}
+
+    {#- Inject Thinking token at the very top of the FIRST system turn -#}
+    {%- if enable_thinking is defined and enable_thinking -%}
+        {{- '<|think|>\n' -}}
+        {%- set ns.prev_message_type = 'think' -%}
+    {%- endif -%}
+
+    {%- if messages[0]['role'] in ['system', 'developer'] -%}
+        {{- messages[0]['content'] | trim -}}
+        {%- set loop_messages = messages[1:] -%}
+    {%- endif -%}
+
+    {%- if tools -%}
+        {%- for tool in tools %}
+            {{- '<|tool>' -}}
+            {{- format_function_declaration(tool) | trim -}}
+            {{- '<tool|>' -}}
+        {%- endfor %}
+        {%- set ns.prev_message_type = 'tool' -%}
+    {%- endif -%}
+
+    {{- '<turn|>\n' -}}
+{%- endif %}
+
+{#- Pre-scan: find last user message index for reasoning guard -#}
+{%- set ns_turn = namespace(last_user_idx=-1) -%}
+{%- for i in range(loop_messages | length) -%}
+    {%- if loop_messages[i]['role'] == 'user' -%}
+        {%- set ns_turn.last_user_idx = i -%}
+    {%- endif -%}
+{%- endfor -%}
+
+{#- Loop through messages -#}
+{%- for message in loop_messages -%}
+    {%- if message['role'] != 'tool' -%}
+    {%- set ns.prev_message_type = None -%}
+    {%- set role = 'model' if message['role'] == 'assistant' else message['role'] -%}
+    {#- Detect continuation: suppress duplicate <|turn>model when previous non-tool message was also assistant -#}
+    {%- set prev_nt = namespace(role=None, found=false) -%}
+    {%- if loop.index0 > 0 -%}
+        {%- for j in range(loop.index0 - 1, -1, -1) -%}
+            {%- if not prev_nt.found -%}
+                {%- if loop_messages[j]['role'] != 'tool' -%}
+                    {%- set prev_nt.role = loop_messages[j]['role'] -%}
+                    {%- set prev_nt.found = true -%}
+                {%- endif -%}
+            {%- endif -%}
+        {%- endfor -%}
+    {%- endif -%}
+    {%- set continue_same_model_turn = (role == 'model' and prev_nt.role == 'assistant') -%}
+    {%- if not continue_same_model_turn -%}
+        {{- '<|turn>' + role + '\n' }}
+    {%- endif -%}
+
+    {#- Render reasoning/reasoning_content as thinking channel -#}
+    {%- set thinking_text = message.get('reasoning') or message.get('reasoning_content') -%}
+    {%- if thinking_text and loop.index0 > ns_turn.last_user_idx and message.get('tool_calls') -%}
+        {{- '<|channel>thought\n' + thinking_text + '\n<channel|>' -}}
+    {%- endif -%}
+
+            {%- if message['tool_calls'] -%}
+                {%- for tool_call in message['tool_calls'] -%}
+                    {%- set function = tool_call['function'] -%}
+                    {{- '<|tool_call>call:' + function['name'] + '{' -}}
+                    {%- if function['arguments'] is mapping -%}
+                        {%- set ns_args = namespace(found_first=false) -%}
+                        {%- for key, value in function['arguments'] | dictsort -%}
+                            {%- if ns_args.found_first %},{% endif -%}
+                            {%- set ns_args.found_first = true -%}
+                            {{- key -}}:{{- format_argument(value, escape_keys=False) -}}
+                        {%- endfor -%}
+                    {%- elif function['arguments'] is string -%}
+                        {{- function['arguments'] -}}
+                    {%- endif -%}
+                    {{- '}<tool_call|>' -}}
+                {%- endfor -%}
+                {%- set ns.prev_message_type = 'tool_call' -%}
+            {%- endif -%}
+
+            {%- set ns_tr_out = namespace(flag=false) -%}
+            {%- if message.get('tool_responses') -%}
+                {#- Legacy: tool_responses embedded on the assistant message (Google/Gemma native) -#}
+                {%- for tool_response in message['tool_responses'] -%}
+                    {{- format_tool_response_block(tool_response['name'] | default('unknown'), tool_response['response']) -}}
+                    {%- set ns_tr_out.flag = true -%}
+                    {%- set ns.prev_message_type = 'tool_response' -%}
+                {%- endfor -%}
+            {%- elif message.get('tool_calls') -%}
+                {#- OpenAI Chat Completions: forward-scan consecutive role:tool messages -#}
+                {%- set ns_tool_scan = namespace(stopped=false) -%}
+                {%- for k in range(loop.index0 + 1, loop_messages | length) -%}
+                    {%- if ns_tool_scan.stopped -%}
+                    {%- elif loop_messages[k]['role'] != 'tool' -%}
+                        {%- set ns_tool_scan.stopped = true -%}
+                    {%- else -%}
+                        {%- set follow = loop_messages[k] -%}
+                        {#- Resolve tool_call_id to function name -#}
+                        {%- set ns_tname = namespace(name=follow.get('name') | default('unknown')) -%}
+                        {%- for tc in message['tool_calls'] -%}
+                            {%- if tc.get('id') == follow.get('tool_call_id') -%}
+                                {%- set ns_tname.name = tc['function']['name'] -%}
+                            {%- endif -%}
+                        {%- endfor -%}
+                        {#- Handle content as string or content-parts array -#}
+                        {%- set tool_body = follow.get('content') -%}
+                        {%- if tool_body is string -%}
+                            {{- format_tool_response_block(ns_tname.name, tool_body) -}}
+                        {%- elif tool_body is sequence and tool_body is not string -%}
+                            {%- set ns_txt = namespace(s='') -%}
+                            {%- for part in tool_body -%}
+                                {%- if part.get('type') == 'text' -%}
+                                    {%- set ns_txt.s = ns_txt.s + (part.get('text') | default('')) -%}
+                                {%- endif -%}
+                            {%- endfor -%}
+                            {{- format_tool_response_block(ns_tname.name, ns_txt.s) -}}
+                        {%- else -%}
+                            {{- format_tool_response_block(ns_tname.name, tool_body) -}}
+                        {%- endif -%}
+                        {%- set ns_tr_out.flag = true -%}
+                        {%- set ns.prev_message_type = 'tool_response' -%}
+                    {%- endif -%}
+                {%- endfor -%}
+            {%- endif -%}
+
+            {%- if message['content'] is string -%}
+                {%- if role == 'model' -%}
+                    {{- strip_thinking(message['content']) -}}
+                {%- else -%}
+                    {{- message['content'] | trim -}}
+                {%- endif -%}
+            {%- elif message['content'] is sequence -%}
+                {%- for item in message['content'] -%}
+                    {%- if item['type'] == 'text' -%}
+                        {%- if role == 'model' -%}
+                            {{- strip_thinking(item['text']) -}}
+                        {%- else -%}
+                            {{- item['text'] | trim -}}
+                        {%- endif -%}
+                    {%- elif item['type'] == 'image' -%}
+                        {{- '<|image|>' -}}
+                        {%- set ns.prev_message_type = 'image' -%}
+                    {%- elif item['type'] == 'audio' -%}
+                        {{- '<|audio|>' -}}
+                        {%- set ns.prev_message_type = 'audio' -%}
+                    {%- elif item['type'] == 'video' -%}
+                        {{- '<|video|>' -}}
+                        {%- set ns.prev_message_type = 'video' -%}
+                    {%- endif -%}
+                {%- endfor -%}
+            {%- endif -%}
+
+        {%- if ns.prev_message_type == 'tool_call' and not ns_tr_out.flag -%}
+            {{- '<|tool_response>' -}}
+        {%- elif not (ns_tr_out.flag and not message.get('content')) -%}
+            {{- '<turn|>\n' -}}
+        {%- endif -%}
+    {%- endif -%}
+{%- endfor -%}
+
+{%- if add_generation_prompt -%}
+    {%- if ns.prev_message_type != 'tool_response' and ns.prev_message_type != 'tool_call' -%}
+        {{- '<|turn>model\n' -}}
+    {%- endif -%}
+{%- endif -%}

checkpoint-200/optimizer.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5c9a141f6be155d771bea07fa6ffa898d3ac626d3ae5ccebb65a6adaae9ce276
+size 75236437

checkpoint-200/processor_config.json

@@ -0,0 +1,75 @@
+{
+  "audio_ms_per_token": 40,
+  "audio_seq_length": 750,
+  "feature_extractor": {
+    "dither": 0.0,
+    "feature_extractor_type": "Gemma4AudioFeatureExtractor",
+    "feature_size": 128,
+    "fft_length": 512,
+    "fft_overdrive": false,
+    "frame_length": 320,
+    "hop_length": 160,
+    "input_scale_factor": 1.0,
+    "max_frequency": 8000.0,
+    "mel_floor": 0.001,
+    "min_frequency": 0.0,
+    "padding_side": "left",
+    "padding_value": 0.0,
+    "per_bin_mean": null,
+    "per_bin_stddev": null,
+    "preemphasis": 0.0,
+    "preemphasis_htk_flavor": true,
+    "return_attention_mask": true,
+    "sampling_rate": 16000
+  },
+  "image_processor": {
+    "do_convert_rgb": true,
+    "do_normalize": false,
+    "do_rescale": true,
+    "do_resize": true,
+    "image_mean": [
+      0.0,
+      0.0,
+      0.0
+    ],
+    "image_processor_type": "Gemma4ImageProcessor",
+    "image_seq_length": 280,
+    "image_std": [
+      1.0,
+      1.0,
+      1.0
+    ],
+    "max_soft_tokens": 280,
+    "patch_size": 16,
+    "pooling_kernel_size": 3,
+    "resample": 3,
+    "rescale_factor": 0.00392156862745098
+  },
+  "image_seq_length": 280,
+  "processor_class": "Gemma4Processor",
+  "video_processor": {
+    "do_convert_rgb": true,
+    "do_normalize": true,
+    "do_rescale": true,
+    "do_resize": true,
+    "do_sample_frames": true,
+    "image_mean": [
+      0.0,
+      0.0,
+      0.0
+    ],
+    "image_std": [
+      1.0,
+      1.0,
+      1.0
+    ],
+    "max_soft_tokens": 70,
+    "num_frames": 32,
+    "patch_size": 16,
+    "pooling_kernel_size": 3,
+    "resample": 3,
+    "rescale_factor": 0.00392156862745098,
+    "return_metadata": false,
+    "video_processor_type": "Gemma4VideoProcessor"
+  }
+}

checkpoint-200/rng_state.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f1d565802a8e26c4e8a31328752b7a7fdc186d9401aa008e65697d0ad8c22e33
+size 14645

checkpoint-200/scheduler.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b018f6bf6da9cbcd24154267bd4a94fb94f45f816081ef434edb910f3a576ce0
+size 1465

checkpoint-200/tokenizer.json

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:cc8d3a0ce36466ccc1278bf987df5f71db1719b9ca6b4118264f45cb627bfe0f
+size 32169626

checkpoint-200/tokenizer_config.json

@@ -0,0 +1,95 @@
+{
+  "audio_token": "<|audio|>",
+  "backend": "tokenizers",
+  "boa_token": "<|audio>",
+  "boi_token": "<|image>",
+  "bos_token": "<bos>",
+  "eoa_token": "<audio|>",
+  "eoc_token": "<channel|>",
+  "eoi_token": "<image|>",
+  "eos_token": "<turn|>",
+  "eot_token": "<turn|>",
+  "escape_token": "<|\"|>",
+  "etc_token": "<tool_call|>",
+  "etd_token": "<tool|>",
+  "etr_token": "<tool_response|>",
+  "extra_special_tokens": [
+    "<|video|>"
+  ],
+  "image_token": "<|image|>",
+  "is_local": false,
+  "mask_token": "<mask>",
+  "model_max_length": 131072,
+  "model_specific_special_tokens": {
+    "audio_token": "<|audio|>",
+    "boa_token": "<|audio>",
+    "boi_token": "<|image>",
+    "eoa_token": "<audio|>",
+    "eoc_token": "<channel|>",
+    "eoi_token": "<image|>",
+    "eot_token": "<turn|>",
+    "escape_token": "<|\"|>",
+    "etc_token": "<tool_call|>",
+    "etd_token": "<tool|>",
+    "etr_token": "<tool_response|>",
+    "image_token": "<|image|>",
+    "soc_token": "<|channel>",
+    "sot_token": "<|turn>",
+    "stc_token": "<|tool_call>",
+    "std_token": "<|tool>",
+    "str_token": "<|tool_response>",
+    "think_token": "<|think|>"
+  },
+  "pad_token": "<pad>",
+  "padding_side": "right",
+  "processor_class": "Gemma4Processor",
+  "response_schema": {
+    "properties": {
+      "content": {
+        "type": "string"
+      },
+      "role": {
+        "const": "assistant"
+      },
+      "thinking": {
+        "type": "string"
+      },
+      "tool_calls": {
+        "items": {
+          "properties": {
+            "function": {
+              "properties": {
+                "arguments": {
+                  "additionalProperties": {},
+                  "type": "object",
+                  "x-parser": "gemma4-tool-call"
+                },
+                "name": {
+                  "type": "string"
+                }
+              },
+              "type": "object",
+              "x-regex": "call\\:(?P<name>\\w+)(?P<arguments>\\{.*\\})"
+            },
+            "type": {
+              "const": "function"
+            }
+          },
+          "type": "object"
+        },
+        "type": "array",
+        "x-regex-iterator": "<\\|tool_call>(.*?)<tool_call\\|>"
+      }
+    },
+    "type": "object",
+    "x-regex": "(\\<\\|channel\\>thought\\n(?P<thinking>.*?)\\<channel\\|\\>)?(?P<tool_calls>\\<\\|tool_call\\>.*\\<tool_call\\|\\>)?(?P<content>(?:(?!\\<turn\\|\\>)(?!\\<\\|tool_response\\>).)+)?(?:\\<turn\\|\\>|\\<\\|tool_response\\>)?"
+  },
+  "soc_token": "<|channel>",
+  "sot_token": "<|turn>",
+  "stc_token": "<|tool_call>",
+  "std_token": "<|tool>",
+  "str_token": "<|tool_response>",
+  "think_token": "<|think|>",
+  "tokenizer_class": "GemmaTokenizer",
+  "unk_token": "<unk>"
+}

checkpoint-200/trainer_state.json

@@ -0,0 +1,2034 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.39980009995002497,
+  "eval_steps": 201,
+  "global_step": 200,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.001999000499750125,
+      "grad_norm": 7.331668376922607,
+      "learning_rate": 0.0,
+      "loss": 5.560573577880859,
+      "num_input_tokens_seen": 16384,
+      "step": 1,
+      "train_runtime": 47.3306,
+      "train_tokens_per_second": 346.161
+    },
+    {
+      "epoch": 0.00399800099950025,
+      "grad_norm": 6.79426908493042,
+      "learning_rate": 4e-05,
+      "loss": 5.523191452026367,
+      "num_input_tokens_seen": 32768,
+      "step": 2,
+      "train_runtime": 57.9053,
+      "train_tokens_per_second": 565.89
+    },
+    {
+      "epoch": 0.005997001499250375,
+      "grad_norm": 6.676287651062012,
+      "learning_rate": 8e-05,
+      "loss": 5.496252536773682,
+      "num_input_tokens_seen": 49152,
+      "step": 3,
+      "train_runtime": 68.4941,
+      "train_tokens_per_second": 717.609
+    },
+    {
+      "epoch": 0.0079960019990005,
+      "grad_norm": 6.097739219665527,
+      "learning_rate": 0.00012,
+      "loss": 5.170515060424805,
+      "num_input_tokens_seen": 65536,
+      "step": 4,
+      "train_runtime": 79.1923,
+      "train_tokens_per_second": 827.555
+    },
+    {
+      "epoch": 0.009995002498750625,
+      "grad_norm": 4.8083930015563965,
+      "learning_rate": 0.00016,
+      "loss": 4.473215579986572,
+      "num_input_tokens_seen": 81920,
+      "step": 5,
+      "train_runtime": 89.8555,
+      "train_tokens_per_second": 911.686
+    },
+    {
+      "epoch": 0.01199400299850075,
+      "grad_norm": 2.681720733642578,
+      "learning_rate": 0.0002,
+      "loss": 3.8302183151245117,
+      "num_input_tokens_seen": 98304,
+      "step": 6,
+      "train_runtime": 100.5028,
+      "train_tokens_per_second": 978.122
+    },
+    {
+      "epoch": 0.013993003498250875,
+      "grad_norm": 1.557854175567627,
+      "learning_rate": 0.00019999950354593194,
+      "loss": 3.418180465698242,
+      "num_input_tokens_seen": 114688,
+      "step": 7,
+      "train_runtime": 111.1678,
+      "train_tokens_per_second": 1031.666
+    },
+    {
+      "epoch": 0.015992003998001,
+      "grad_norm": 1.7741894721984863,
+      "learning_rate": 0.0001999980141886571,
+      "loss": 3.4549918174743652,
+      "num_input_tokens_seen": 131072,
+      "step": 8,
+      "train_runtime": 121.8347,
+      "train_tokens_per_second": 1075.819
+    },
+    {
+      "epoch": 0.017991004497751123,
+      "grad_norm": 1.2810389995574951,
+      "learning_rate": 0.00019999553194296343,
+      "loss": 3.329780340194702,
+      "num_input_tokens_seen": 147456,
+      "step": 9,
+      "train_runtime": 132.4884,
+      "train_tokens_per_second": 1112.973
+    },
+    {
+      "epoch": 0.01999000499750125,
+      "grad_norm": 1.1019361019134521,
+      "learning_rate": 0.00019999205683349732,
+      "loss": 3.2165651321411133,
+      "num_input_tokens_seen": 163840,
+      "step": 10,
+      "train_runtime": 143.1724,
+      "train_tokens_per_second": 1144.355
+    },
+    {
+      "epoch": 0.021989005497251374,
+      "grad_norm": 1.0735076665878296,
+      "learning_rate": 0.00019998758889476346,
+      "loss": 3.1698949337005615,
+      "num_input_tokens_seen": 180224,
+      "step": 11,
+      "train_runtime": 153.8339,
+      "train_tokens_per_second": 1171.549
+    },
+    {
+      "epoch": 0.0239880059970015,
+      "grad_norm": 1.1004042625427246,
+      "learning_rate": 0.00019998212817112437,
+      "loss": 3.0940957069396973,
+      "num_input_tokens_seen": 196608,
+      "step": 12,
+      "train_runtime": 164.5019,
+      "train_tokens_per_second": 1195.171
+    },
+    {
+      "epoch": 0.025987006496751622,
+      "grad_norm": 0.9596648216247559,
+      "learning_rate": 0.00019997567471679998,
+      "loss": 3.059084892272949,
+      "num_input_tokens_seen": 212992,
+      "step": 13,
+      "train_runtime": 175.1844,
+      "train_tokens_per_second": 1215.816
+    },
+    {
+      "epoch": 0.02798600699650175,
+      "grad_norm": 0.8726735711097717,
+      "learning_rate": 0.0001999682285958672,
+      "loss": 2.9257137775421143,
+      "num_input_tokens_seen": 229376,
+      "step": 14,
+      "train_runtime": 185.924,
+      "train_tokens_per_second": 1233.708
+    },
+    {
+      "epoch": 0.029985007496251874,
+      "grad_norm": 0.7122098207473755,
+      "learning_rate": 0.00019995978988225912,
+      "loss": 2.878443717956543,
+      "num_input_tokens_seen": 245760,
+      "step": 15,
+      "train_runtime": 196.6216,
+      "train_tokens_per_second": 1249.913
+    },
+    {
+      "epoch": 0.031984007996002,
+      "grad_norm": 0.7174533009529114,
+      "learning_rate": 0.0001999503586597645,
+      "loss": 2.998406410217285,
+      "num_input_tokens_seen": 262144,
+      "step": 16,
+      "train_runtime": 207.2962,
+      "train_tokens_per_second": 1264.587
+    },
+    {
+      "epoch": 0.03398300849575212,
+      "grad_norm": 0.7693482637405396,
+      "learning_rate": 0.00019993993502202665,
+      "loss": 2.928849935531616,
+      "num_input_tokens_seen": 278528,
+      "step": 17,
+      "train_runtime": 218.0082,
+      "train_tokens_per_second": 1277.603
+    },
+    {
+      "epoch": 0.035982008995502246,
+      "grad_norm": 0.6065346598625183,
+      "learning_rate": 0.00019992851907254275,
+      "loss": 2.991607189178467,
+      "num_input_tokens_seen": 294912,
+      "step": 18,
+      "train_runtime": 228.703,
+      "train_tokens_per_second": 1289.498
+    },
+    {
+      "epoch": 0.03798100949525238,
+      "grad_norm": 0.5639224052429199,
+      "learning_rate": 0.00019991611092466265,
+      "loss": 2.9910218715667725,
+      "num_input_tokens_seen": 311296,
+      "step": 19,
+      "train_runtime": 239.4409,
+      "train_tokens_per_second": 1300.096
+    },
+    {
+      "epoch": 0.0399800099950025,
+      "grad_norm": 0.5850295424461365,
+      "learning_rate": 0.0001999027107015879,
+      "loss": 2.8126330375671387,
+      "num_input_tokens_seen": 327680,
+      "step": 20,
+      "train_runtime": 250.1755,
+      "train_tokens_per_second": 1309.801
+    },
+    {
+      "epoch": 0.041979010494752625,
+      "grad_norm": 0.6322516798973083,
+      "learning_rate": 0.0001998883185363704,
+      "loss": 3.0385286808013916,
+      "num_input_tokens_seen": 344064,
+      "step": 21,
+      "train_runtime": 260.8995,
+      "train_tokens_per_second": 1318.76
+    },
+    {
+      "epoch": 0.04397801099450275,
+      "grad_norm": 0.5834643840789795,
+      "learning_rate": 0.00019987293457191112,
+      "loss": 2.8844664096832275,
+      "num_input_tokens_seen": 360364,
+      "step": 22,
+      "train_runtime": 271.6189,
+      "train_tokens_per_second": 1326.726
+    },
+    {
+      "epoch": 0.04597701149425287,
+      "grad_norm": 0.656398594379425,
+      "learning_rate": 0.00019985655896095867,
+      "loss": 2.738499879837036,
+      "num_input_tokens_seen": 376596,
+      "step": 23,
+      "train_runtime": 282.2862,
+      "train_tokens_per_second": 1334.093
+    },
+    {
+      "epoch": 0.047976011994003,
+      "grad_norm": 0.590424120426178,
+      "learning_rate": 0.00019983919186610788,
+      "loss": 2.7959787845611572,
+      "num_input_tokens_seen": 392924,
+      "step": 24,
+      "train_runtime": 293.0197,
+      "train_tokens_per_second": 1340.947
+    },
+    {
+      "epoch": 0.04997501249375312,
+      "grad_norm": 0.5563395619392395,
+      "learning_rate": 0.000199820833459798,
+      "loss": 2.889263868331909,
+      "num_input_tokens_seen": 409308,
+      "step": 25,
+      "train_runtime": 303.7202,
+      "train_tokens_per_second": 1347.648
+    },
+    {
+      "epoch": 0.051974012993503245,
+      "grad_norm": 0.5884258151054382,
+      "learning_rate": 0.00019980148392431118,
+      "loss": 2.767672300338745,
+      "num_input_tokens_seen": 425692,
+      "step": 26,
+      "train_runtime": 314.4518,
+      "train_tokens_per_second": 1353.759
+    },
+    {
+      "epoch": 0.053973013493253376,
+      "grad_norm": 0.5780461430549622,
+      "learning_rate": 0.00019978114345177049,
+      "loss": 2.6928277015686035,
+      "num_input_tokens_seen": 442076,
+      "step": 27,
+      "train_runtime": 325.1895,
+      "train_tokens_per_second": 1359.441
+    },
+    {
+      "epoch": 0.0559720139930035,
+      "grad_norm": 0.6170737147331238,
+      "learning_rate": 0.00019975981224413817,
+      "loss": 2.664564609527588,
+      "num_input_tokens_seen": 457770,
+      "step": 28,
+      "train_runtime": 335.4736,
+      "train_tokens_per_second": 1364.549
+    },
+    {
+      "epoch": 0.057971014492753624,
+      "grad_norm": 0.548454761505127,
+      "learning_rate": 0.00019973749051321348,
+      "loss": 2.7781982421875,
+      "num_input_tokens_seen": 474154,
+      "step": 29,
+      "train_runtime": 346.2202,
+      "train_tokens_per_second": 1369.516
+    },
+    {
+      "epoch": 0.05997001499250375,
+      "grad_norm": 0.6355477571487427,
+      "learning_rate": 0.00019971417848063073,
+      "loss": 2.7185378074645996,
+      "num_input_tokens_seen": 490260,
+      "step": 30,
+      "train_runtime": 356.748,
+      "train_tokens_per_second": 1374.247
+    },
+    {
+      "epoch": 0.06196901549225387,
+      "grad_norm": 0.6777451634407043,
+      "learning_rate": 0.000199689876377857,
+      "loss": 2.770723342895508,
+      "num_input_tokens_seen": 506644,
+      "step": 31,
+      "train_runtime": 367.484,
+      "train_tokens_per_second": 1378.683
+    },
+    {
+      "epoch": 0.063968015992004,
+      "grad_norm": 0.6335959434509277,
+      "learning_rate": 0.0001996645844461898,
+      "loss": 2.907606601715088,
+      "num_input_tokens_seen": 523028,
+      "step": 32,
+      "train_runtime": 378.2014,
+      "train_tokens_per_second": 1382.935
+    },
+    {
+      "epoch": 0.06596701649175413,
+      "grad_norm": 0.6486599445343018,
+      "learning_rate": 0.00019963830293675484,
+      "loss": 2.7316300868988037,
+      "num_input_tokens_seen": 539412,
+      "step": 33,
+      "train_runtime": 388.9274,
+      "train_tokens_per_second": 1386.922
+    },
+    {
+      "epoch": 0.06796601699150424,
+      "grad_norm": 0.5915785431861877,
+      "learning_rate": 0.0001996110321105033,
+      "loss": 2.620595932006836,
+      "num_input_tokens_seen": 555796,
+      "step": 34,
+      "train_runtime": 399.6287,
+      "train_tokens_per_second": 1390.781
+    },
+    {
+      "epoch": 0.06996501749125437,
+      "grad_norm": 0.6603999137878418,
+      "learning_rate": 0.00019958277223820948,
+      "loss": 2.708037853240967,
+      "num_input_tokens_seen": 572180,
+      "step": 35,
+      "train_runtime": 410.3614,
+      "train_tokens_per_second": 1394.332
+    },
+    {
+      "epoch": 0.07196401799100449,
+      "grad_norm": 0.7603235840797424,
+      "learning_rate": 0.00019955352360046793,
+      "loss": 2.644076108932495,
+      "num_input_tokens_seen": 588564,
+      "step": 36,
+      "train_runtime": 421.0633,
+      "train_tokens_per_second": 1397.804
+    },
+    {
+      "epoch": 0.07396301849075462,
+      "grad_norm": 0.6619148254394531,
+      "learning_rate": 0.00019952328648769076,
+      "loss": 2.5154783725738525,
+      "num_input_tokens_seen": 604948,
+      "step": 37,
+      "train_runtime": 431.7586,
+      "train_tokens_per_second": 1401.125
+    },
+    {
+      "epoch": 0.07596201899050475,
+      "grad_norm": 0.622572124004364,
+      "learning_rate": 0.00019949206120010472,
+      "loss": 2.5505049228668213,
+      "num_input_tokens_seen": 621332,
+      "step": 38,
+      "train_runtime": 442.4992,
+      "train_tokens_per_second": 1404.143
+    },
+    {
+      "epoch": 0.07796101949025487,
+      "grad_norm": 0.6390085816383362,
+      "learning_rate": 0.00019945984804774824,
+      "loss": 2.5269813537597656,
+      "num_input_tokens_seen": 637716,
+      "step": 39,
+      "train_runtime": 453.1913,
+      "train_tokens_per_second": 1407.167
+    },
+    {
+      "epoch": 0.079960019990005,
+      "grad_norm": 0.5684909224510193,
+      "learning_rate": 0.0001994266473504683,
+      "loss": 2.6473617553710938,
+      "num_input_tokens_seen": 654100,
+      "step": 40,
+      "train_runtime": 463.9273,
+      "train_tokens_per_second": 1409.919
+    },
+    {
+      "epoch": 0.08195902048975512,
+      "grad_norm": 0.6122294068336487,
+      "learning_rate": 0.00019939245943791738,
+      "loss": 2.7440977096557617,
+      "num_input_tokens_seen": 670484,
+      "step": 41,
+      "train_runtime": 474.661,
+      "train_tokens_per_second": 1412.553
+    },
+    {
+      "epoch": 0.08395802098950525,
+      "grad_norm": 0.5501320958137512,
+      "learning_rate": 0.00019935728464955002,
+      "loss": 2.5517232418060303,
+      "num_input_tokens_seen": 686868,
+      "step": 42,
+      "train_runtime": 485.3586,
+      "train_tokens_per_second": 1415.176
+    },
+    {
+      "epoch": 0.08595702148925537,
+      "grad_norm": 0.6636989712715149,
+      "learning_rate": 0.0001993211233346195,
+      "loss": 2.4997594356536865,
+      "num_input_tokens_seen": 703250,
+      "step": 43,
+      "train_runtime": 496.0626,
+      "train_tokens_per_second": 1417.664
+    },
+    {
+      "epoch": 0.0879560219890055,
+      "grad_norm": 0.6815392971038818,
+      "learning_rate": 0.00019928397585217456,
+      "loss": 2.722917079925537,
+      "num_input_tokens_seen": 719634,
+      "step": 44,
+      "train_runtime": 506.766,
+      "train_tokens_per_second": 1420.052
+    },
+    {
+      "epoch": 0.08995502248875563,
+      "grad_norm": 0.6484745144844055,
+      "learning_rate": 0.0001992458425710555,
+      "loss": 2.6333179473876953,
+      "num_input_tokens_seen": 736018,
+      "step": 45,
+      "train_runtime": 517.4891,
+      "train_tokens_per_second": 1422.287
+    },
+    {
+      "epoch": 0.09195402298850575,
+      "grad_norm": 0.5829582810401917,
+      "learning_rate": 0.0001992067238698908,
+      "loss": 2.548511266708374,
+      "num_input_tokens_seen": 752308,
+      "step": 46,
+      "train_runtime": 528.2175,
+      "train_tokens_per_second": 1424.239
+    },
+    {
+      "epoch": 0.09395302348825588,
+      "grad_norm": 0.6313190460205078,
+      "learning_rate": 0.0001991666201370932,
+      "loss": 2.6169910430908203,
+      "num_input_tokens_seen": 768692,
+      "step": 47,
+      "train_runtime": 538.9658,
+      "train_tokens_per_second": 1426.235
+    },
+    {
+      "epoch": 0.095952023988006,
+      "grad_norm": 0.6699250340461731,
+      "learning_rate": 0.00019912553177085598,
+      "loss": 2.595137357711792,
+      "num_input_tokens_seen": 785046,
+      "step": 48,
+      "train_runtime": 549.724,
+      "train_tokens_per_second": 1428.073
+    },
+    {
+      "epoch": 0.09795102448775612,
+      "grad_norm": 0.7118059396743774,
+      "learning_rate": 0.0001990834591791488,
+      "loss": 2.652780055999756,
+      "num_input_tokens_seen": 801430,
+      "step": 49,
+      "train_runtime": 560.4265,
+      "train_tokens_per_second": 1430.036
+    },
+    {
+      "epoch": 0.09995002498750624,
+      "grad_norm": 0.5986912250518799,
+      "learning_rate": 0.0001990404027797139,
+      "loss": 2.656085968017578,
+      "num_input_tokens_seen": 817814,
+      "step": 50,
+      "train_runtime": 571.1351,
+      "train_tokens_per_second": 1431.91
+    },
+    {
+      "epoch": 0.10194902548725637,
+      "grad_norm": 0.75722336769104,
+      "learning_rate": 0.00019899636300006174,
+      "loss": 2.543043613433838,
+      "num_input_tokens_seen": 834194,
+      "step": 51,
+      "train_runtime": 581.8781,
+      "train_tokens_per_second": 1433.623
+    },
+    {
+      "epoch": 0.10394802598700649,
+      "grad_norm": 0.6150996685028076,
+      "learning_rate": 0.00019895134027746693,
+      "loss": 2.7043724060058594,
+      "num_input_tokens_seen": 850578,
+      "step": 52,
+      "train_runtime": 592.6296,
+      "train_tokens_per_second": 1435.261
+    },
+    {
+      "epoch": 0.10594702648675662,
+      "grad_norm": 0.6528828144073486,
+      "learning_rate": 0.00019890533505896368,
+      "loss": 2.6164627075195312,
+      "num_input_tokens_seen": 866962,
+      "step": 53,
+      "train_runtime": 603.3564,
+      "train_tokens_per_second": 1436.899
+    },
+    {
+      "epoch": 0.10794602698650675,
+      "grad_norm": 0.6227877140045166,
+      "learning_rate": 0.00019885834780134157,
+      "loss": 2.379403591156006,
+      "num_input_tokens_seen": 883346,
+      "step": 54,
+      "train_runtime": 614.0739,
+      "train_tokens_per_second": 1438.501
+    },
+    {
+      "epoch": 0.10994502748625687,
+      "grad_norm": 0.6063210368156433,
+      "learning_rate": 0.00019881037897114093,
+      "loss": 2.456653118133545,
+      "num_input_tokens_seen": 899730,
+      "step": 55,
+      "train_runtime": 624.7836,
+      "train_tokens_per_second": 1440.067
+    },
+    {
+      "epoch": 0.111944027986007,
+      "grad_norm": 0.5717106461524963,
+      "learning_rate": 0.00019876142904464813,
+      "loss": 2.566009283065796,
+      "num_input_tokens_seen": 916114,
+      "step": 56,
+      "train_runtime": 635.519,
+      "train_tokens_per_second": 1441.521
+    },
+    {
+      "epoch": 0.11394302848575712,
+      "grad_norm": 0.6836978793144226,
+      "learning_rate": 0.00019871149850789102,
+      "loss": 2.4524693489074707,
+      "num_input_tokens_seen": 932498,
+      "step": 57,
+      "train_runtime": 646.2281,
+      "train_tokens_per_second": 1442.986
+    },
+    {
+      "epoch": 0.11594202898550725,
+      "grad_norm": 0.6186497211456299,
+      "learning_rate": 0.00019866058785663392,
+      "loss": 2.41436767578125,
+      "num_input_tokens_seen": 948882,
+      "step": 58,
+      "train_runtime": 656.9409,
+      "train_tokens_per_second": 1444.395
+    },
+    {
+      "epoch": 0.11794102948525736,
+      "grad_norm": 0.6606503129005432,
+      "learning_rate": 0.00019860869759637289,
+      "loss": 2.5905051231384277,
+      "num_input_tokens_seen": 965266,
+      "step": 59,
+      "train_runtime": 667.6704,
+      "train_tokens_per_second": 1445.722
+    },
+    {
+      "epoch": 0.1199400299850075,
+      "grad_norm": 0.6339024901390076,
+      "learning_rate": 0.00019855582824233049,
+      "loss": 2.4514667987823486,
+      "num_input_tokens_seen": 981650,
+      "step": 60,
+      "train_runtime": 678.3831,
+      "train_tokens_per_second": 1447.044
+    },
+    {
+      "epoch": 0.12193903048475763,
+      "grad_norm": 0.6130167841911316,
+      "learning_rate": 0.00019850198031945082,
+      "loss": 2.602674722671509,
+      "num_input_tokens_seen": 998034,
+      "step": 61,
+      "train_runtime": 689.1504,
+      "train_tokens_per_second": 1448.209
+    },
+    {
+      "epoch": 0.12393803098450774,
+      "grad_norm": 0.6347913146018982,
+      "learning_rate": 0.00019844715436239437,
+      "loss": 2.6927785873413086,
+      "num_input_tokens_seen": 1014130,
+      "step": 62,
+      "train_runtime": 699.6977,
+      "train_tokens_per_second": 1449.383
+    },
+    {
+      "epoch": 0.12593703148425786,
+      "grad_norm": 0.6667070388793945,
+      "learning_rate": 0.00019839135091553244,
+      "loss": 2.3476438522338867,
+      "num_input_tokens_seen": 1030514,
+      "step": 63,
+      "train_runtime": 710.4257,
+      "train_tokens_per_second": 1450.558
+    },
+    {
+      "epoch": 0.127936031984008,
+      "grad_norm": 0.7157728672027588,
+      "learning_rate": 0.00019833457053294206,
+      "loss": 2.543039321899414,
+      "num_input_tokens_seen": 1046898,
+      "step": 64,
+      "train_runtime": 721.1477,
+      "train_tokens_per_second": 1451.711
+    },
+    {
+      "epoch": 0.12993503248375812,
+      "grad_norm": 0.6451475024223328,
+      "learning_rate": 0.0001982768137784002,
+      "loss": 2.5445199012756348,
+      "num_input_tokens_seen": 1063282,
+      "step": 65,
+      "train_runtime": 731.841,
+      "train_tokens_per_second": 1452.887
+    },
+    {
+      "epoch": 0.13193403298350825,
+      "grad_norm": 0.7809988856315613,
+      "learning_rate": 0.00019821808122537848,
+      "loss": 2.509563446044922,
+      "num_input_tokens_seen": 1079666,
+      "step": 66,
+      "train_runtime": 742.6017,
+      "train_tokens_per_second": 1453.896
+    },
+    {
+      "epoch": 0.13393303348325838,
+      "grad_norm": 0.7288188338279724,
+      "learning_rate": 0.0001981583734570371,
+      "loss": 2.5006520748138428,
+      "num_input_tokens_seen": 1096050,
+      "step": 67,
+      "train_runtime": 753.333,
+      "train_tokens_per_second": 1454.934
+    },
+    {
+      "epoch": 0.1359320339830085,
+      "grad_norm": 0.6310102343559265,
+      "learning_rate": 0.00019809769106621938,
+      "loss": 2.5584566593170166,
+      "num_input_tokens_seen": 1112434,
+      "step": 68,
+      "train_runtime": 764.0483,
+      "train_tokens_per_second": 1455.973
+    },
+    {
+      "epoch": 0.13793103448275862,
+      "grad_norm": 0.6697754263877869,
+      "learning_rate": 0.00019803603465544574,
+      "loss": 2.4173929691314697,
+      "num_input_tokens_seen": 1128818,
+      "step": 69,
+      "train_runtime": 774.7452,
+      "train_tokens_per_second": 1457.018
+    },
+    {
+      "epoch": 0.13993003498250875,
+      "grad_norm": 0.6117935180664062,
+      "learning_rate": 0.00019797340483690765,
+      "loss": 2.5029242038726807,
+      "num_input_tokens_seen": 1145202,
+      "step": 70,
+      "train_runtime": 785.4566,
+      "train_tokens_per_second": 1458.008
+    },
+    {
+      "epoch": 0.14192903548225888,
+      "grad_norm": 0.5620390772819519,
+      "learning_rate": 0.0001979098022324617,
+      "loss": 2.590005397796631,
+      "num_input_tokens_seen": 1161586,
+      "step": 71,
+      "train_runtime": 796.1852,
+      "train_tokens_per_second": 1458.939
+    },
+    {
+      "epoch": 0.14392803598200898,
+      "grad_norm": 0.7210037112236023,
+      "learning_rate": 0.00019784522747362337,
+      "loss": 2.4453177452087402,
+      "num_input_tokens_seen": 1177970,
+      "step": 72,
+      "train_runtime": 806.9126,
+      "train_tokens_per_second": 1459.848
+    },
+    {
+      "epoch": 0.14592703648175911,
+      "grad_norm": 0.6530371308326721,
+      "learning_rate": 0.0001977796812015606,
+      "loss": 2.5644259452819824,
+      "num_input_tokens_seen": 1194354,
+      "step": 73,
+      "train_runtime": 817.6329,
+      "train_tokens_per_second": 1460.746
+    },
+    {
+      "epoch": 0.14792603698150925,
+      "grad_norm": 0.5794245004653931,
+      "learning_rate": 0.00019771316406708777,
+      "loss": 2.452650547027588,
+      "num_input_tokens_seen": 1210738,
+      "step": 74,
+      "train_runtime": 828.3384,
+      "train_tokens_per_second": 1461.647
+    },
+    {
+      "epoch": 0.14992503748125938,
+      "grad_norm": 0.6062936186790466,
+      "learning_rate": 0.00019764567673065882,
+      "loss": 2.3821656703948975,
+      "num_input_tokens_seen": 1227122,
+      "step": 75,
+      "train_runtime": 839.0713,
+      "train_tokens_per_second": 1462.476
+    },
+    {
+      "epoch": 0.1519240379810095,
+      "grad_norm": 0.6602532863616943,
+      "learning_rate": 0.00019757721986236103,
+      "loss": 2.5787909030914307,
+      "num_input_tokens_seen": 1243506,
+      "step": 76,
+      "train_runtime": 849.7678,
+      "train_tokens_per_second": 1463.348
+    },
+    {
+      "epoch": 0.1539230384807596,
+      "grad_norm": 0.5799870491027832,
+      "learning_rate": 0.00019750779414190824,
+      "loss": 2.4096055030822754,
+      "num_input_tokens_seen": 1259890,
+      "step": 77,
+      "train_runtime": 860.4648,
+      "train_tokens_per_second": 1464.197
+    },
+    {
+      "epoch": 0.15592203898050974,
+      "grad_norm": 0.6456904411315918,
+      "learning_rate": 0.0001974374002586341,
+      "loss": 2.3753206729888916,
+      "num_input_tokens_seen": 1276274,
+      "step": 78,
+      "train_runtime": 871.1761,
+      "train_tokens_per_second": 1465.001
+    },
+    {
+      "epoch": 0.15792103948025987,
+      "grad_norm": 0.6789486408233643,
+      "learning_rate": 0.00019736603891148513,
+      "loss": 2.556735038757324,
+      "num_input_tokens_seen": 1292658,
+      "step": 79,
+      "train_runtime": 881.92,
+      "train_tokens_per_second": 1465.732
+    },
+    {
+      "epoch": 0.15992003998001,
+      "grad_norm": 0.5870137810707092,
+      "learning_rate": 0.00019729371080901402,
+      "loss": 2.5232064723968506,
+      "num_input_tokens_seen": 1309042,
+      "step": 80,
+      "train_runtime": 892.6343,
+      "train_tokens_per_second": 1466.493
+    },
+    {
+      "epoch": 0.1619190404797601,
+      "grad_norm": 0.5439413189888,
+      "learning_rate": 0.00019722041666937233,
+      "loss": 2.518394947052002,
+      "num_input_tokens_seen": 1325426,
+      "step": 81,
+      "train_runtime": 903.3552,
+      "train_tokens_per_second": 1467.226
+    },
+    {
+      "epoch": 0.16391804097951024,
+      "grad_norm": 0.7084252834320068,
+      "learning_rate": 0.00019714615722030357,
+      "loss": 2.5006942749023438,
+      "num_input_tokens_seen": 1341810,
+      "step": 82,
+      "train_runtime": 914.0655,
+      "train_tokens_per_second": 1467.958
+    },
+    {
+      "epoch": 0.16591704147926037,
+      "grad_norm": 0.6800078749656677,
+      "learning_rate": 0.00019707093319913587,
+      "loss": 2.5847601890563965,
+      "num_input_tokens_seen": 1358194,
+      "step": 83,
+      "train_runtime": 924.7859,
+      "train_tokens_per_second": 1468.658
+    },
+    {
+      "epoch": 0.1679160419790105,
+      "grad_norm": 0.5962506532669067,
+      "learning_rate": 0.0001969947453527746,
+      "loss": 2.394505023956299,
+      "num_input_tokens_seen": 1374578,
+      "step": 84,
+      "train_runtime": 935.4976,
+      "train_tokens_per_second": 1469.355
+    },
+    {
+      "epoch": 0.16991504247876063,
+      "grad_norm": 0.659408688545227,
+      "learning_rate": 0.00019691759443769512,
+      "loss": 2.5332651138305664,
+      "num_input_tokens_seen": 1390962,
+      "step": 85,
+      "train_runtime": 946.2179,
+      "train_tokens_per_second": 1470.023
+    },
+    {
+      "epoch": 0.17191404297851073,
+      "grad_norm": 0.7143354415893555,
+      "learning_rate": 0.00019683948121993516,
+      "loss": 2.618718147277832,
+      "num_input_tokens_seen": 1407346,
+      "step": 86,
+      "train_runtime": 956.9405,
+      "train_tokens_per_second": 1470.672
+    },
+    {
+      "epoch": 0.17391304347826086,
+      "grad_norm": 0.6828016638755798,
+      "learning_rate": 0.00019676040647508718,
+      "loss": 2.506481170654297,
+      "num_input_tokens_seen": 1423730,
+      "step": 87,
+      "train_runtime": 967.6653,
+      "train_tokens_per_second": 1471.304
+    },
+    {
+      "epoch": 0.175912043978011,
+      "grad_norm": 0.6377484202384949,
+      "learning_rate": 0.00019668037098829078,
+      "loss": 2.5501859188079834,
+      "num_input_tokens_seen": 1440114,
+      "step": 88,
+      "train_runtime": 978.4125,
+      "train_tokens_per_second": 1471.888
+    },
+    {
+      "epoch": 0.17791104447776113,
+      "grad_norm": 0.5860212445259094,
+      "learning_rate": 0.00019659937555422476,
+      "loss": 2.428554058074951,
+      "num_input_tokens_seen": 1456258,
+      "step": 89,
+      "train_runtime": 989.0475,
+      "train_tokens_per_second": 1472.384
+    },
+    {
+      "epoch": 0.17991004497751126,
+      "grad_norm": 0.6303457021713257,
+      "learning_rate": 0.00019651742097709946,
+      "loss": 2.508391857147217,
+      "num_input_tokens_seen": 1472642,
+      "step": 90,
+      "train_runtime": 999.7602,
+      "train_tokens_per_second": 1472.995
+    },
+    {
+      "epoch": 0.18190904547726136,
+      "grad_norm": 0.6128749847412109,
+      "learning_rate": 0.0001964345080706485,
+      "loss": 2.3652591705322266,
+      "num_input_tokens_seen": 1489026,
+      "step": 91,
+      "train_runtime": 1010.4805,
+      "train_tokens_per_second": 1473.582
+    },
+    {
+      "epoch": 0.1839080459770115,
+      "grad_norm": 0.5624603033065796,
+      "learning_rate": 0.00019635063765812091,
+      "loss": 2.4202356338500977,
+      "num_input_tokens_seen": 1505410,
+      "step": 92,
+      "train_runtime": 1021.1725,
+      "train_tokens_per_second": 1474.198
+    },
+    {
+      "epoch": 0.18590704647676162,
+      "grad_norm": 0.586696207523346,
+      "learning_rate": 0.0001962658105722728,
+      "loss": 2.507248640060425,
+      "num_input_tokens_seen": 1521794,
+      "step": 93,
+      "train_runtime": 1031.8815,
+      "train_tokens_per_second": 1474.776
+    },
+    {
+      "epoch": 0.18790604697651175,
+      "grad_norm": 0.6841021180152893,
+      "learning_rate": 0.00019618002765535919,
+      "loss": 2.434886932373047,
+      "num_input_tokens_seen": 1538178,
+      "step": 94,
+      "train_runtime": 1042.6122,
+      "train_tokens_per_second": 1475.312
+    },
+    {
+      "epoch": 0.18990504747626186,
+      "grad_norm": 0.6928579807281494,
+      "learning_rate": 0.00019609328975912568,
+      "loss": 2.5311965942382812,
+      "num_input_tokens_seen": 1554562,
+      "step": 95,
+      "train_runtime": 1053.3189,
+      "train_tokens_per_second": 1475.87
+    },
+    {
+      "epoch": 0.191904047976012,
+      "grad_norm": 0.6460245251655579,
+      "learning_rate": 0.00019600559774479988,
+      "loss": 2.2193102836608887,
+      "num_input_tokens_seen": 1570880,
+      "step": 96,
+      "train_runtime": 1063.9799,
+      "train_tokens_per_second": 1476.419
+    },
+    {
+      "epoch": 0.19390304847576212,
+      "grad_norm": 0.6225016117095947,
+      "learning_rate": 0.00019591695248308296,
+      "loss": 2.555119514465332,
+      "num_input_tokens_seen": 1587264,
+      "step": 97,
+      "train_runtime": 1074.6849,
+      "train_tokens_per_second": 1476.958
+    },
+    {
+      "epoch": 0.19590204897551225,
+      "grad_norm": 0.6219892501831055,
+      "learning_rate": 0.00019582735485414092,
+      "loss": 2.430931329727173,
+      "num_input_tokens_seen": 1603236,
+      "step": 98,
+      "train_runtime": 1085.1012,
+      "train_tokens_per_second": 1477.499
+    },
+    {
+      "epoch": 0.19790104947526238,
+      "grad_norm": 0.5690893530845642,
+      "learning_rate": 0.0001957368057475959,
+      "loss": 2.392733097076416,
+      "num_input_tokens_seen": 1619564,
+      "step": 99,
+      "train_runtime": 1095.7816,
+      "train_tokens_per_second": 1477.999
+    },
+    {
+      "epoch": 0.19990004997501248,
+      "grad_norm": 0.6479955911636353,
+      "learning_rate": 0.00019564530606251736,
+      "loss": 2.5462844371795654,
+      "num_input_tokens_seen": 1635948,
+      "step": 100,
+      "train_runtime": 1106.4822,
+      "train_tokens_per_second": 1478.513
+    },
+    {
+      "epoch": 0.20189905047476261,
+      "grad_norm": 0.657764732837677,
+      "learning_rate": 0.00019555285670741311,
+      "loss": 2.51003098487854,
+      "num_input_tokens_seen": 1652332,
+      "step": 101,
+      "train_runtime": 1117.1814,
+      "train_tokens_per_second": 1479.018
+    },
+    {
+      "epoch": 0.20389805097451275,
+      "grad_norm": 0.6311238408088684,
+      "learning_rate": 0.0001954594586002203,
+      "loss": 2.3784544467926025,
+      "num_input_tokens_seen": 1668716,
+      "step": 102,
+      "train_runtime": 1127.8625,
+      "train_tokens_per_second": 1479.538
+    },
+    {
+      "epoch": 0.20589705147426288,
+      "grad_norm": 0.6337195038795471,
+      "learning_rate": 0.00019536511266829637,
+      "loss": 2.335280656814575,
+      "num_input_tokens_seen": 1685100,
+      "step": 103,
+      "train_runtime": 1138.5584,
+      "train_tokens_per_second": 1480.03
+    },
+    {
+      "epoch": 0.20789605197401298,
+      "grad_norm": 0.7382025122642517,
+      "learning_rate": 0.00019526981984840976,
+      "loss": 2.535940170288086,
+      "num_input_tokens_seen": 1701484,
+      "step": 104,
+      "train_runtime": 1149.2375,
+      "train_tokens_per_second": 1480.533
+    },
+    {
+      "epoch": 0.2098950524737631,
+      "grad_norm": 0.650765597820282,
+      "learning_rate": 0.00019517358108673056,
+      "loss": 2.451458215713501,
+      "num_input_tokens_seen": 1717868,
+      "step": 105,
+      "train_runtime": 1159.9206,
+      "train_tokens_per_second": 1481.022
+    },
+    {
+      "epoch": 0.21189405297351324,
+      "grad_norm": 0.5573831796646118,
+      "learning_rate": 0.00019507639733882135,
+      "loss": 2.317359447479248,
+      "num_input_tokens_seen": 1734252,
+      "step": 106,
+      "train_runtime": 1170.6077,
+      "train_tokens_per_second": 1481.497
+    },
+    {
+      "epoch": 0.21389305347326337,
+      "grad_norm": 0.5417583584785461,
+      "learning_rate": 0.00019497826956962745,
+      "loss": 2.3837292194366455,
+      "num_input_tokens_seen": 1750636,
+      "step": 107,
+      "train_runtime": 1181.3713,
+      "train_tokens_per_second": 1481.868
+    },
+    {
+      "epoch": 0.2158920539730135,
+      "grad_norm": 0.5987706780433655,
+      "learning_rate": 0.0001948791987534674,
+      "loss": 2.3685407638549805,
+      "num_input_tokens_seen": 1767020,
+      "step": 108,
+      "train_runtime": 1192.0855,
+      "train_tokens_per_second": 1482.293
+    },
+    {
+      "epoch": 0.2178910544727636,
+      "grad_norm": 0.6503360271453857,
+      "learning_rate": 0.0001947791858740235,
+      "loss": 2.3225173950195312,
+      "num_input_tokens_seen": 1783404,
+      "step": 109,
+      "train_runtime": 1202.8358,
+      "train_tokens_per_second": 1482.666
+    },
+    {
+      "epoch": 0.21989005497251374,
+      "grad_norm": 0.642593502998352,
+      "learning_rate": 0.00019467823192433165,
+      "loss": 2.2902991771698,
+      "num_input_tokens_seen": 1799788,
+      "step": 110,
+      "train_runtime": 1213.5615,
+      "train_tokens_per_second": 1483.063
+    },
+    {
+      "epoch": 0.22188905547226387,
+      "grad_norm": 0.6346986293792725,
+      "learning_rate": 0.00019457633790677194,
+      "loss": 2.3180410861968994,
+      "num_input_tokens_seen": 1816172,
+      "step": 111,
+      "train_runtime": 1224.2315,
+      "train_tokens_per_second": 1483.52
+    },
+    {
+      "epoch": 0.223888055972014,
+      "grad_norm": 0.6983447074890137,
+      "learning_rate": 0.00019447350483305827,
+      "loss": 2.346426248550415,
+      "num_input_tokens_seen": 1832556,
+      "step": 112,
+      "train_runtime": 1234.9117,
+      "train_tokens_per_second": 1483.957
+    },
+    {
+      "epoch": 0.22588705647176413,
+      "grad_norm": 0.6771863698959351,
+      "learning_rate": 0.00019436973372422868,
+      "loss": 2.393000841140747,
+      "num_input_tokens_seen": 1848940,
+      "step": 113,
+      "train_runtime": 1245.7103,
+      "train_tokens_per_second": 1484.246
+    },
+    {
+      "epoch": 0.22788605697151423,
+      "grad_norm": 0.6571068167686462,
+      "learning_rate": 0.0001942650256106349,
+      "loss": 2.609287738800049,
+      "num_input_tokens_seen": 1865324,
+      "step": 114,
+      "train_runtime": 1256.4874,
+      "train_tokens_per_second": 1484.554
+    },
+    {
+      "epoch": 0.22988505747126436,
+      "grad_norm": 0.6709936857223511,
+      "learning_rate": 0.0001941593815319323,
+      "loss": 2.2739598751068115,
+      "num_input_tokens_seen": 1881708,
+      "step": 115,
+      "train_runtime": 1267.267,
+      "train_tokens_per_second": 1484.855
+    },
+    {
+      "epoch": 0.2318840579710145,
+      "grad_norm": 0.5464649796485901,
+      "learning_rate": 0.00019405280253706955,
+      "loss": 2.2550160884857178,
+      "num_input_tokens_seen": 1898092,
+      "step": 116,
+      "train_runtime": 1278.0428,
+      "train_tokens_per_second": 1485.155
+    },
+    {
+      "epoch": 0.23388305847076463,
+      "grad_norm": 0.7197510600090027,
+      "learning_rate": 0.00019394528968427817,
+      "loss": 2.490752696990967,
+      "num_input_tokens_seen": 1914476,
+      "step": 117,
+      "train_runtime": 1288.8266,
+      "train_tokens_per_second": 1485.441
+    },
+    {
+      "epoch": 0.23588205897051473,
+      "grad_norm": 0.7368646860122681,
+      "learning_rate": 0.00019383684404106202,
+      "loss": 2.4146125316619873,
+      "num_input_tokens_seen": 1930378,
+      "step": 118,
+      "train_runtime": 1299.2727,
+      "train_tokens_per_second": 1485.737
+    },
+    {
+      "epoch": 0.23788105947026486,
+      "grad_norm": 0.6332247853279114,
+      "learning_rate": 0.00019372746668418672,
+      "loss": 2.4597740173339844,
+      "num_input_tokens_seen": 1946762,
+      "step": 119,
+      "train_runtime": 1309.9725,
+      "train_tokens_per_second": 1486.109
+    },
+    {
+      "epoch": 0.239880059970015,
+      "grad_norm": 0.6263355016708374,
+      "learning_rate": 0.00019361715869966893,
+      "loss": 2.322871208190918,
+      "num_input_tokens_seen": 1963146,
+      "step": 120,
+      "train_runtime": 1320.6856,
+      "train_tokens_per_second": 1486.46
+    },
+    {
+      "epoch": 0.24187906046976512,
+      "grad_norm": 0.6257721185684204,
+      "learning_rate": 0.00019350592118276558,
+      "loss": 2.417241334915161,
+      "num_input_tokens_seen": 1979530,
+      "step": 121,
+      "train_runtime": 1331.4948,
+      "train_tokens_per_second": 1486.697
+    },
+    {
+      "epoch": 0.24387806096951525,
+      "grad_norm": 0.6563782691955566,
+      "learning_rate": 0.00019339375523796308,
+      "loss": 2.389177083969116,
+      "num_input_tokens_seen": 1995914,
+      "step": 122,
+      "train_runtime": 1342.5386,
+      "train_tokens_per_second": 1486.672
+    },
+    {
+      "epoch": 0.24587706146926536,
+      "grad_norm": 0.672744631767273,
+      "learning_rate": 0.0001932806619789662,
+      "loss": 2.3840231895446777,
+      "num_input_tokens_seen": 2012298,
+      "step": 123,
+      "train_runtime": 1353.2371,
+      "train_tokens_per_second": 1487.025
+    },
+    {
+      "epoch": 0.2478760619690155,
+      "grad_norm": 0.696761965751648,
+      "learning_rate": 0.00019316664252868712,
+      "loss": 2.441990852355957,
+      "num_input_tokens_seen": 2028682,
+      "step": 124,
+      "train_runtime": 1363.927,
+      "train_tokens_per_second": 1487.383
+    },
+    {
+      "epoch": 0.24987506246876562,
+      "grad_norm": 0.6644465327262878,
+      "learning_rate": 0.0001930516980192342,
+      "loss": 2.3033244609832764,
+      "num_input_tokens_seen": 2045066,
+      "step": 125,
+      "train_runtime": 1374.6254,
+      "train_tokens_per_second": 1487.726
+    },
+    {
+      "epoch": 0.2518740629685157,
+      "grad_norm": 0.6689712405204773,
+      "learning_rate": 0.00019293582959190085,
+      "loss": 2.4898993968963623,
+      "num_input_tokens_seen": 2061450,
+      "step": 126,
+      "train_runtime": 1385.3167,
+      "train_tokens_per_second": 1488.071
+    },
+    {
+      "epoch": 0.2538730634682659,
+      "grad_norm": 0.66810542345047,
+      "learning_rate": 0.0001928190383971541,
+      "loss": 2.326140880584717,
+      "num_input_tokens_seen": 2077834,
+      "step": 127,
+      "train_runtime": 1396.0051,
+      "train_tokens_per_second": 1488.414
+    },
+    {
+      "epoch": 0.255872063968016,
+      "grad_norm": 0.6487640738487244,
+      "learning_rate": 0.00019270132559462327,
+      "loss": 2.2935826778411865,
+      "num_input_tokens_seen": 2093808,
+      "step": 128,
+      "train_runtime": 1406.4513,
+      "train_tokens_per_second": 1488.717
+    },
+    {
+      "epoch": 0.25787106446776614,
+      "grad_norm": 0.6936726570129395,
+      "learning_rate": 0.0001925826923530883,
+      "loss": 2.269087314605713,
+      "num_input_tokens_seen": 2110192,
+      "step": 129,
+      "train_runtime": 1417.1429,
+      "train_tokens_per_second": 1489.047
+    },
+    {
+      "epoch": 0.25987006496751625,
+      "grad_norm": 0.5996974110603333,
+      "learning_rate": 0.00019246313985046834,
+      "loss": 2.4819657802581787,
+      "num_input_tokens_seen": 2126576,
+      "step": 130,
+      "train_runtime": 1427.9474,
+      "train_tokens_per_second": 1489.254
+    },
+    {
+      "epoch": 0.26186906546726635,
+      "grad_norm": 0.6297034025192261,
+      "learning_rate": 0.00019234266927380987,
+      "loss": 2.38704252243042,
+      "num_input_tokens_seen": 2142960,
+      "step": 131,
+      "train_runtime": 1438.7434,
+      "train_tokens_per_second": 1489.466
+    },
+    {
+      "epoch": 0.2638680659670165,
+      "grad_norm": 0.7511478066444397,
+      "learning_rate": 0.0001922212818192751,
+      "loss": 2.4105327129364014,
+      "num_input_tokens_seen": 2159344,
+      "step": 132,
+      "train_runtime": 1449.6524,
+      "train_tokens_per_second": 1489.56
+    },
+    {
+      "epoch": 0.2658670664667666,
+      "grad_norm": 0.6854334473609924,
+      "learning_rate": 0.0001920989786921299,
+      "loss": 2.398444414138794,
+      "num_input_tokens_seen": 2175728,
+      "step": 133,
+      "train_runtime": 1460.519,
+      "train_tokens_per_second": 1489.695
+    },
+    {
+      "epoch": 0.26786606696651677,
+      "grad_norm": 0.6412373185157776,
+      "learning_rate": 0.00019197576110673196,
+      "loss": 2.313157320022583,
+      "num_input_tokens_seen": 2192084,
+      "step": 134,
+      "train_runtime": 1471.2447,
+      "train_tokens_per_second": 1489.952
+    },
+    {
+      "epoch": 0.2698650674662669,
+      "grad_norm": 0.7024304270744324,
+      "learning_rate": 0.00019185163028651877,
+      "loss": 2.4681296348571777,
+      "num_input_tokens_seen": 2208468,
+      "step": 135,
+      "train_runtime": 1481.9499,
+      "train_tokens_per_second": 1490.245
+    },
+    {
+      "epoch": 0.271864067966017,
+      "grad_norm": 0.7034132480621338,
+      "learning_rate": 0.0001917265874639953,
+      "loss": 2.445939064025879,
+      "num_input_tokens_seen": 2224852,
+      "step": 136,
+      "train_runtime": 1492.7366,
+      "train_tokens_per_second": 1490.452
+    },
+    {
+      "epoch": 0.27386306846576713,
+      "grad_norm": 0.6865590810775757,
+      "learning_rate": 0.00019160063388072191,
+      "loss": 2.463561773300171,
+      "num_input_tokens_seen": 2241236,
+      "step": 137,
+      "train_runtime": 1503.6505,
+      "train_tokens_per_second": 1490.53
+    },
+    {
+      "epoch": 0.27586206896551724,
+      "grad_norm": 0.6259234547615051,
+      "learning_rate": 0.000191473770787302,
+      "loss": 2.3744146823883057,
+      "num_input_tokens_seen": 2257620,
+      "step": 138,
+      "train_runtime": 1514.3626,
+      "train_tokens_per_second": 1490.805
+    },
+    {
+      "epoch": 0.27786106946526734,
+      "grad_norm": 0.6290014982223511,
+      "learning_rate": 0.0001913459994433695,
+      "loss": 2.335517168045044,
+      "num_input_tokens_seen": 2274004,
+      "step": 139,
+      "train_runtime": 1525.3057,
+      "train_tokens_per_second": 1490.851
+    },
+    {
+      "epoch": 0.2798600699650175,
+      "grad_norm": 0.7093214988708496,
+      "learning_rate": 0.00019121732111757652,
+      "loss": 2.437976598739624,
+      "num_input_tokens_seen": 2290388,
+      "step": 140,
+      "train_runtime": 1536.4965,
+      "train_tokens_per_second": 1490.656
+    },
+    {
+      "epoch": 0.2818590704647676,
+      "grad_norm": 0.6410729885101318,
+      "learning_rate": 0.00019108773708758064,
+      "loss": 2.268831729888916,
+      "num_input_tokens_seen": 2306772,
+      "step": 141,
+      "train_runtime": 1547.2457,
+      "train_tokens_per_second": 1490.889
+    },
+    {
+      "epoch": 0.28385807096451776,
+      "grad_norm": 0.6548078656196594,
+      "learning_rate": 0.00019095724864003218,
+      "loss": 2.300246000289917,
+      "num_input_tokens_seen": 2323156,
+      "step": 142,
+      "train_runtime": 1557.9309,
+      "train_tokens_per_second": 1491.18
+    },
+    {
+      "epoch": 0.28585707146426786,
+      "grad_norm": 0.6462287306785583,
+      "learning_rate": 0.00019082585707056157,
+      "loss": 2.25825834274292,
+      "num_input_tokens_seen": 2339510,
+      "step": 143,
+      "train_runtime": 1568.9863,
+      "train_tokens_per_second": 1491.097
+    },
+    {
+      "epoch": 0.28785607196401797,
+      "grad_norm": 0.6960925459861755,
+      "learning_rate": 0.00019069356368376646,
+      "loss": 2.2979907989501953,
+      "num_input_tokens_seen": 2355894,
+      "step": 144,
+      "train_runtime": 1579.7309,
+      "train_tokens_per_second": 1491.326
+    },
+    {
+      "epoch": 0.2898550724637681,
+      "grad_norm": 0.59915691614151,
+      "learning_rate": 0.00019056036979319858,
+      "loss": 2.4597928524017334,
+      "num_input_tokens_seen": 2372278,
+      "step": 145,
+      "train_runtime": 1590.4017,
+      "train_tokens_per_second": 1491.622
+    },
+    {
+      "epoch": 0.29185407296351823,
+      "grad_norm": 0.72845458984375,
+      "learning_rate": 0.00019042627672135088,
+      "loss": 2.38205623626709,
+      "num_input_tokens_seen": 2388662,
+      "step": 146,
+      "train_runtime": 1601.1154,
+      "train_tokens_per_second": 1491.874
+    },
+    {
+      "epoch": 0.2938530734632684,
+      "grad_norm": 0.6576253771781921,
+      "learning_rate": 0.0001902912857996445,
+      "loss": 2.3961737155914307,
+      "num_input_tokens_seen": 2405046,
+      "step": 147,
+      "train_runtime": 1611.8141,
+      "train_tokens_per_second": 1492.136
+    },
+    {
+      "epoch": 0.2958520739630185,
+      "grad_norm": 0.7284627556800842,
+      "learning_rate": 0.00019015539836841517,
+      "loss": 2.388429641723633,
+      "num_input_tokens_seen": 2421430,
+      "step": 148,
+      "train_runtime": 1622.5054,
+      "train_tokens_per_second": 1492.402
+    },
+    {
+      "epoch": 0.2978510744627686,
+      "grad_norm": 0.6298946142196655,
+      "learning_rate": 0.00019001861577690028,
+      "loss": 2.337505340576172,
+      "num_input_tokens_seen": 2437814,
+      "step": 149,
+      "train_runtime": 1633.2149,
+      "train_tokens_per_second": 1492.647
+    },
+    {
+      "epoch": 0.29985007496251875,
+      "grad_norm": 0.6117843985557556,
+      "learning_rate": 0.00018988093938322533,
+      "loss": 2.470449924468994,
+      "num_input_tokens_seen": 2454198,
+      "step": 150,
+      "train_runtime": 1643.9612,
+      "train_tokens_per_second": 1492.856
+    },
+    {
+      "epoch": 0.30184907546226886,
+      "grad_norm": 0.7032175660133362,
+      "learning_rate": 0.00018974237055439044,
+      "loss": 2.420793294906616,
+      "num_input_tokens_seen": 2470582,
+      "step": 151,
+      "train_runtime": 1654.6573,
+      "train_tokens_per_second": 1493.108
+    },
+    {
+      "epoch": 0.303848075962019,
+      "grad_norm": 1.0140293836593628,
+      "learning_rate": 0.00018960291066625674,
+      "loss": 2.314863681793213,
+      "num_input_tokens_seen": 2486966,
+      "step": 152,
+      "train_runtime": 1666.4585,
+      "train_tokens_per_second": 1492.366
+    },
+    {
+      "epoch": 0.3058470764617691,
+      "grad_norm": 0.6635705232620239,
+      "learning_rate": 0.00018946256110353286,
+      "loss": 2.4007279872894287,
+      "num_input_tokens_seen": 2503350,
+      "step": 153,
+      "train_runtime": 1677.1869,
+      "train_tokens_per_second": 1492.589
+    },
+    {
+      "epoch": 0.3078460769615192,
+      "grad_norm": 0.6424167156219482,
+      "learning_rate": 0.000189321323259761,
+      "loss": 2.3012707233428955,
+      "num_input_tokens_seen": 2519470,
+      "step": 154,
+      "train_runtime": 1687.6968,
+      "train_tokens_per_second": 1492.845
+    },
+    {
+      "epoch": 0.3098450774612694,
+      "grad_norm": 0.5892760157585144,
+      "learning_rate": 0.00018917919853730323,
+      "loss": 2.2420153617858887,
+      "num_input_tokens_seen": 2535002,
+      "step": 155,
+      "train_runtime": 1697.8558,
+      "train_tokens_per_second": 1493.061
+    },
+    {
+      "epoch": 0.3118440779610195,
+      "grad_norm": 0.5827703475952148,
+      "learning_rate": 0.0001890361883473274,
+      "loss": 2.320924997329712,
+      "num_input_tokens_seen": 2551386,
+      "step": 156,
+      "train_runtime": 1708.542,
+      "train_tokens_per_second": 1493.312
+    },
+    {
+      "epoch": 0.31384307846076964,
+      "grad_norm": 0.5837633609771729,
+      "learning_rate": 0.00018889229410979342,
+      "loss": 2.333136558532715,
+      "num_input_tokens_seen": 2567770,
+      "step": 157,
+      "train_runtime": 1719.2276,
+      "train_tokens_per_second": 1493.56
+    },
+    {
+      "epoch": 0.31584207896051975,
+      "grad_norm": 0.6836022734642029,
+      "learning_rate": 0.00018874751725343885,
+      "loss": 2.394927501678467,
+      "num_input_tokens_seen": 2584154,
+      "step": 158,
+      "train_runtime": 1729.916,
+      "train_tokens_per_second": 1493.803
+    },
+    {
+      "epoch": 0.31784107946026985,
+      "grad_norm": 0.6186059713363647,
+      "learning_rate": 0.00018860185921576483,
+      "loss": 2.296104669570923,
+      "num_input_tokens_seen": 2600500,
+      "step": 159,
+      "train_runtime": 1740.6466,
+      "train_tokens_per_second": 1493.985
+    },
+    {
+      "epoch": 0.31984007996002,
+      "grad_norm": 0.6260522603988647,
+      "learning_rate": 0.00018845532144302194,
+      "loss": 2.4206161499023438,
+      "num_input_tokens_seen": 2616884,
+      "step": 160,
+      "train_runtime": 1751.3314,
+      "train_tokens_per_second": 1494.225
+    },
+    {
+      "epoch": 0.3218390804597701,
+      "grad_norm": 0.7028151154518127,
+      "learning_rate": 0.0001883079053901956,
+      "loss": 2.3263039588928223,
+      "num_input_tokens_seen": 2633268,
+      "step": 161,
+      "train_runtime": 1762.0189,
+      "train_tokens_per_second": 1494.461
+    },
+    {
+      "epoch": 0.3238380809595202,
+      "grad_norm": 0.6303103566169739,
+      "learning_rate": 0.00018815961252099183,
+      "loss": 2.41351580619812,
+      "num_input_tokens_seen": 2649132,
+      "step": 162,
+      "train_runtime": 1772.3051,
+      "train_tokens_per_second": 1494.738
+    },
+    {
+      "epoch": 0.3258370814592704,
+      "grad_norm": 0.6776697039604187,
+      "learning_rate": 0.0001880104443078225,
+      "loss": 2.277194023132324,
+      "num_input_tokens_seen": 2665516,
+      "step": 163,
+      "train_runtime": 1783.0109,
+      "train_tokens_per_second": 1494.952
+    },
+    {
+      "epoch": 0.3278360819590205,
+      "grad_norm": 0.6697080135345459,
+      "learning_rate": 0.00018786040223179107,
+      "loss": 2.486637830734253,
+      "num_input_tokens_seen": 2681900,
+      "step": 164,
+      "train_runtime": 1793.6996,
+      "train_tokens_per_second": 1495.178
+    },
+    {
+      "epoch": 0.32983508245877063,
+      "grad_norm": 0.6267579197883606,
+      "learning_rate": 0.00018770948778267744,
+      "loss": 2.339092254638672,
+      "num_input_tokens_seen": 2698284,
+      "step": 165,
+      "train_runtime": 1804.404,
+      "train_tokens_per_second": 1495.388
+    },
+    {
+      "epoch": 0.33183408295852074,
+      "grad_norm": 0.5976248383522034,
+      "learning_rate": 0.00018755770245892345,
+      "loss": 2.163480520248413,
+      "num_input_tokens_seen": 2714668,
+      "step": 166,
+      "train_runtime": 1815.0936,
+      "train_tokens_per_second": 1495.608
+    },
+    {
+      "epoch": 0.33383308345827084,
+      "grad_norm": 0.6570451259613037,
+      "learning_rate": 0.00018740504776761796,
+      "loss": 2.56025767326355,
+      "num_input_tokens_seen": 2731052,
+      "step": 167,
+      "train_runtime": 1825.7808,
+      "train_tokens_per_second": 1495.827
+    },
+    {
+      "epoch": 0.335832083958021,
+      "grad_norm": 0.7304461002349854,
+      "learning_rate": 0.0001872515252244818,
+      "loss": 2.5441439151763916,
+      "num_input_tokens_seen": 2747436,
+      "step": 168,
+      "train_runtime": 1836.4678,
+      "train_tokens_per_second": 1496.044
+    },
+    {
+      "epoch": 0.3378310844577711,
+      "grad_norm": 0.5998393893241882,
+      "learning_rate": 0.00018709713635385282,
+      "loss": 2.444791316986084,
+      "num_input_tokens_seen": 2763820,
+      "step": 169,
+      "train_runtime": 1847.1609,
+      "train_tokens_per_second": 1496.253
+    },
+    {
+      "epoch": 0.33983008495752126,
+      "grad_norm": 0.6098015904426575,
+      "learning_rate": 0.00018694188268867062,
+      "loss": 2.60428786277771,
+      "num_input_tokens_seen": 2780204,
+      "step": 170,
+      "train_runtime": 1857.8534,
+      "train_tokens_per_second": 1496.46
+    },
+    {
+      "epoch": 0.34182908545727136,
+      "grad_norm": 0.647736668586731,
+      "learning_rate": 0.00018678576577046158,
+      "loss": 2.439578056335449,
+      "num_input_tokens_seen": 2796588,
+      "step": 171,
+      "train_runtime": 1868.601,
+      "train_tokens_per_second": 1496.621
+    },
+    {
+      "epoch": 0.34382808595702147,
+      "grad_norm": 0.6672942638397217,
+      "learning_rate": 0.00018662878714932314,
+      "loss": 2.454664707183838,
+      "num_input_tokens_seen": 2812972,
+      "step": 172,
+      "train_runtime": 1879.3469,
+      "train_tokens_per_second": 1496.782
+    },
+    {
+      "epoch": 0.3458270864567716,
+      "grad_norm": 0.5663049817085266,
+      "learning_rate": 0.00018647094838390892,
+      "loss": 2.4026777744293213,
+      "num_input_tokens_seen": 2829356,
+      "step": 173,
+      "train_runtime": 1890.0635,
+      "train_tokens_per_second": 1496.963
+    },
+    {
+      "epoch": 0.34782608695652173,
+      "grad_norm": 0.7377136945724487,
+      "learning_rate": 0.00018631225104141278,
+      "loss": 2.3012824058532715,
+      "num_input_tokens_seen": 2845740,
+      "step": 174,
+      "train_runtime": 1900.7835,
+      "train_tokens_per_second": 1497.141
+    },
+    {
+      "epoch": 0.3498250874562719,
+      "grad_norm": 0.5479752421379089,
+      "learning_rate": 0.00018615269669755363,
+      "loss": 2.2582085132598877,
+      "num_input_tokens_seen": 2862124,
+      "step": 175,
+      "train_runtime": 1911.4641,
+      "train_tokens_per_second": 1497.346
+    },
+    {
+      "epoch": 0.351824087956022,
+      "grad_norm": 0.6023082137107849,
+      "learning_rate": 0.00018599228693655942,
+      "loss": 2.313683032989502,
+      "num_input_tokens_seen": 2878508,
+      "step": 176,
+      "train_runtime": 1922.1605,
+      "train_tokens_per_second": 1497.538
+    },
+    {
+      "epoch": 0.3538230884557721,
+      "grad_norm": 0.7141623497009277,
+      "learning_rate": 0.00018583102335115187,
+      "loss": 2.5196468830108643,
+      "num_input_tokens_seen": 2894892,
+      "step": 177,
+      "train_runtime": 1932.8615,
+      "train_tokens_per_second": 1497.723
+    },
+    {
+      "epoch": 0.35582208895552225,
+      "grad_norm": 0.687178373336792,
+      "learning_rate": 0.00018566890754253008,
+      "loss": 2.3450679779052734,
+      "num_input_tokens_seen": 2911276,
+      "step": 178,
+      "train_runtime": 1943.5645,
+      "train_tokens_per_second": 1497.905
+    },
+    {
+      "epoch": 0.35782108945527236,
+      "grad_norm": 0.6856889128684998,
+      "learning_rate": 0.0001855059411203552,
+      "loss": 2.3960957527160645,
+      "num_input_tokens_seen": 2927660,
+      "step": 179,
+      "train_runtime": 1954.2507,
+      "train_tokens_per_second": 1498.099
+    },
+    {
+      "epoch": 0.3598200899550225,
+      "grad_norm": 0.6607837080955505,
+      "learning_rate": 0.00018534212570273407,
+      "loss": 2.223076343536377,
+      "num_input_tokens_seen": 2943420,
+      "step": 180,
+      "train_runtime": 1964.4892,
+      "train_tokens_per_second": 1498.313
+    },
+    {
+      "epoch": 0.3618190904547726,
+      "grad_norm": 0.6468161344528198,
+      "learning_rate": 0.0001851774629162033,
+      "loss": 2.3109865188598633,
+      "num_input_tokens_seen": 2959804,
+      "step": 181,
+      "train_runtime": 1975.1656,
+      "train_tokens_per_second": 1498.509
+    },
+    {
+      "epoch": 0.3638180909545227,
+      "grad_norm": 0.6777626276016235,
+      "learning_rate": 0.00018501195439571313,
+      "loss": 2.2806036472320557,
+      "num_input_tokens_seen": 2976188,
+      "step": 182,
+      "train_runtime": 1985.8207,
+      "train_tokens_per_second": 1498.719
+    },
+    {
+      "epoch": 0.3658170914542729,
+      "grad_norm": 0.6999329328536987,
+      "learning_rate": 0.00018484560178461104,
+      "loss": 2.4783687591552734,
+      "num_input_tokens_seen": 2992028,
+      "step": 183,
+      "train_runtime": 1996.1274,
+      "train_tokens_per_second": 1498.916
+    },
+    {
+      "epoch": 0.367816091954023,
+      "grad_norm": 0.6081796884536743,
+      "learning_rate": 0.0001846784067346257,
+      "loss": 2.398671865463257,
+      "num_input_tokens_seen": 3008412,
+      "step": 184,
+      "train_runtime": 2006.8397,
+      "train_tokens_per_second": 1499.079
+    },
+    {
+      "epoch": 0.3698150924537731,
+      "grad_norm": 0.6715173125267029,
+      "learning_rate": 0.0001845103709058504,
+      "loss": 2.1319808959960938,
+      "num_input_tokens_seen": 3024796,
+      "step": 185,
+      "train_runtime": 2017.4994,
+      "train_tokens_per_second": 1499.28
+    },
+    {
+      "epoch": 0.37181409295352325,
+      "grad_norm": 0.7541316747665405,
+      "learning_rate": 0.00018434149596672649,
+      "loss": 2.4389028549194336,
+      "num_input_tokens_seen": 3041180,
+      "step": 186,
+      "train_runtime": 2028.2192,
+      "train_tokens_per_second": 1499.434
+    },
+    {
+      "epoch": 0.37381309345327335,
+      "grad_norm": 0.6820785999298096,
+      "learning_rate": 0.000184171783594027,
+      "loss": 2.481621742248535,
+      "num_input_tokens_seen": 3057564,
+      "step": 187,
+      "train_runtime": 2039.1238,
+      "train_tokens_per_second": 1499.45
+    },
+    {
+      "epoch": 0.3758120939530235,
+      "grad_norm": 0.7726727724075317,
+      "learning_rate": 0.0001840012354728399,
+      "loss": 2.523444652557373,
+      "num_input_tokens_seen": 3073948,
+      "step": 188,
+      "train_runtime": 2050.0699,
+      "train_tokens_per_second": 1499.436
+    },
+    {
+      "epoch": 0.3778110944527736,
+      "grad_norm": 0.6423644423484802,
+      "learning_rate": 0.0001838298532965513,
+      "loss": 2.2019622325897217,
+      "num_input_tokens_seen": 3090332,
+      "step": 189,
+      "train_runtime": 2061.039,
+      "train_tokens_per_second": 1499.405
+    },
+    {
+      "epoch": 0.3798100949525237,
+      "grad_norm": 0.5699455738067627,
+      "learning_rate": 0.0001836576387668289,
+      "loss": 2.1509408950805664,
+      "num_input_tokens_seen": 3106716,
+      "step": 190,
+      "train_runtime": 2071.8862,
+      "train_tokens_per_second": 1499.463
+    },
+    {
+      "epoch": 0.3818090954522739,
+      "grad_norm": 0.6167863607406616,
+      "learning_rate": 0.00018348459359360467,
+      "loss": 2.287490129470825,
+      "num_input_tokens_seen": 3123100,
+      "step": 191,
+      "train_runtime": 2082.637,
+      "train_tokens_per_second": 1499.589
+    },
+    {
+      "epoch": 0.383808095952024,
+      "grad_norm": 0.7306831479072571,
+      "learning_rate": 0.0001833107194950582,
+      "loss": 2.3720479011535645,
+      "num_input_tokens_seen": 3139484,
+      "step": 192,
+      "train_runtime": 2093.3405,
+      "train_tokens_per_second": 1499.748
+    },
+    {
+      "epoch": 0.38580709645177413,
+      "grad_norm": 0.6145524978637695,
+      "learning_rate": 0.00018313601819759965,
+      "loss": 2.202296257019043,
+      "num_input_tokens_seen": 3155868,
+      "step": 193,
+      "train_runtime": 2104.0509,
+      "train_tokens_per_second": 1499.901
+    },
+    {
+      "epoch": 0.38780609695152424,
+      "grad_norm": 0.809516191482544,
+      "learning_rate": 0.00018296049143585238,
+      "loss": 2.2910845279693604,
+      "num_input_tokens_seen": 3172126,
+      "step": 194,
+      "train_runtime": 2114.868,
+      "train_tokens_per_second": 1499.917
+    },
+    {
+      "epoch": 0.38980509745127434,
+      "grad_norm": 0.6672027111053467,
+      "learning_rate": 0.00018278414095263587,
+      "loss": 2.3636233806610107,
+      "num_input_tokens_seen": 3188510,
+      "step": 195,
+      "train_runtime": 2125.6081,
+      "train_tokens_per_second": 1500.046
+    },
+    {
+      "epoch": 0.3918040979510245,
+      "grad_norm": 0.5965846180915833,
+      "learning_rate": 0.0001826069684989484,
+      "loss": 2.3528826236724854,
+      "num_input_tokens_seen": 3204894,
+      "step": 196,
+      "train_runtime": 2136.3136,
+      "train_tokens_per_second": 1500.198
+    },
+    {
+      "epoch": 0.3938030984507746,
+      "grad_norm": 0.613410234451294,
+      "learning_rate": 0.00018242897583394975,
+      "loss": 2.2517004013061523,
+      "num_input_tokens_seen": 3221278,
+      "step": 197,
+      "train_runtime": 2147.0319,
+      "train_tokens_per_second": 1500.34
+    },
+    {
+      "epoch": 0.39580209895052476,
+      "grad_norm": 0.6586496829986572,
+      "learning_rate": 0.00018225016472494347,
+      "loss": 2.3239808082580566,
+      "num_input_tokens_seen": 3237662,
+      "step": 198,
+      "train_runtime": 2157.7134,
+      "train_tokens_per_second": 1500.506
+    },
+    {
+      "epoch": 0.39780109945027486,
+      "grad_norm": 0.6816010475158691,
+      "learning_rate": 0.0001820705369473597,
+      "loss": 2.1372361183166504,
+      "num_input_tokens_seen": 3254046,
+      "step": 199,
+      "train_runtime": 2168.4076,
+      "train_tokens_per_second": 1500.662
+    },
+    {
+      "epoch": 0.39980009995002497,
+      "grad_norm": 0.6197704672813416,
+      "learning_rate": 0.0001818900942847372,
+      "loss": 2.1767396926879883,
+      "num_input_tokens_seen": 3270430,
+      "step": 200,
+      "train_runtime": 2179.1174,
+      "train_tokens_per_second": 1500.805
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 1002,
+  "num_input_tokens_seen": 3270430,
+  "num_train_epochs": 2,
+  "save_steps": 200,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 8.926110141650496e+16,
+  "train_batch_size": 2,
+  "trial_name": null,
+  "trial_params": null
+}

checkpoint-200/training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ae7a694dd4875ee97139bdd8ebc7349b14ce4d529c119642d14f7f7f5ebd9df4
+size 5777

checkpoint-400/README.md

@@ -0,0 +1,210 @@
+---
+base_model: chenhaodev/unsloth_gemma-4-E4B-medical-o1-reasoning-SFT-merged
+library_name: peft
+pipeline_tag: text-generation
+tags:
+- base_model:adapter:chenhaodev/unsloth_gemma-4-E4B-medical-o1-reasoning-SFT-merged
+- lora
+- sft
+- transformers
+- trl
+- unsloth
+---
+
+# Model Card for Model ID
+
+<!-- Provide a quick summary of what the model is/does. -->
+
+
+
+## Model Details
+
+### Model Description
+
+<!-- Provide a longer summary of what this model is. -->
+
+
+
+- **Developed by:** [More Information Needed]
+- **Funded by [optional]:** [More Information Needed]
+- **Shared by [optional]:** [More Information Needed]
+- **Model type:** [More Information Needed]
+- **Language(s) (NLP):** [More Information Needed]
+- **License:** [More Information Needed]
+- **Finetuned from model [optional]:** [More Information Needed]
+
+### Model Sources [optional]
+
+<!-- Provide the basic links for the model. -->
+
+- **Repository:** [More Information Needed]
+- **Paper [optional]:** [More Information Needed]
+- **Demo [optional]:** [More Information Needed]
+
+## Uses
+
+<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
+
+### Direct Use
+
+<!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
+
+[More Information Needed]
+
+### Downstream Use [optional]
+
+<!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->
+
+[More Information Needed]
+
+### Out-of-Scope Use
+
+<!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
+
+[More Information Needed]
+
+## Bias, Risks, and Limitations
+
+<!-- This section is meant to convey both technical and sociotechnical limitations. -->
+
+[More Information Needed]
+
+### Recommendations
+
+<!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->
+
+Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
+
+## How to Get Started with the Model
+
+Use the code below to get started with the model.
+
+[More Information Needed]
+
+## Training Details
+
+### Training Data
+
+<!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
+
+[More Information Needed]
+
+### Training Procedure
+
+<!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
+
+#### Preprocessing [optional]
+
+[More Information Needed]
+
+
+#### Training Hyperparameters
+
+- **Training regime:** [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->
+
+#### Speeds, Sizes, Times [optional]
+
+<!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
+
+[More Information Needed]
+
+## Evaluation
+
+<!-- This section describes the evaluation protocols and provides the results. -->
+
+### Testing Data, Factors & Metrics
+
+#### Testing Data
+
+<!-- This should link to a Dataset Card if possible. -->
+
+[More Information Needed]
+
+#### Factors
+
+<!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->
+
+[More Information Needed]
+
+#### Metrics
+
+<!-- These are the evaluation metrics being used, ideally with a description of why. -->
+
+[More Information Needed]
+
+### Results
+
+[More Information Needed]
+
+#### Summary
+
+
+
+## Model Examination [optional]
+
+<!-- Relevant interpretability work for the model goes here -->
+
+[More Information Needed]
+
+## Environmental Impact
+
+<!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly -->
+
+Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700).
+
+- **Hardware Type:** [More Information Needed]
+- **Hours used:** [More Information Needed]
+- **Cloud Provider:** [More Information Needed]
+- **Compute Region:** [More Information Needed]
+- **Carbon Emitted:** [More Information Needed]
+
+## Technical Specifications [optional]
+
+### Model Architecture and Objective
+
+[More Information Needed]
+
+### Compute Infrastructure
+
+[More Information Needed]
+
+#### Hardware
+
+[More Information Needed]
+
+#### Software
+
+[More Information Needed]
+
+## Citation [optional]
+
+<!-- If there is a paper or blog post introducing the model, the APA and Bibtex information for that should go in this section. -->
+
+**BibTeX:**
+
+[More Information Needed]
+
+**APA:**
+
+[More Information Needed]
+
+## Glossary [optional]
+
+<!-- If relevant, include terms and calculations in this section that can help readers understand the model or model card. -->
+
+[More Information Needed]
+
+## More Information [optional]
+
+[More Information Needed]
+
+## Model Card Authors [optional]
+
+[More Information Needed]
+
+## Model Card Contact
+
+[More Information Needed]
+### Framework versions
+
+- PEFT 0.18.1

checkpoint-400/adapter_config.json

@@ -0,0 +1,50 @@
+{
+  "alora_invocation_tokens": null,
+  "alpha_pattern": {},
+  "arrow_config": null,
+  "auto_mapping": {
+    "base_model_class": "Gemma4ForConditionalGeneration",
+    "parent_library": "transformers.models.gemma4.modeling_gemma4",
+    "unsloth_fixed": true
+  },
+  "base_model_name_or_path": "chenhaodev/unsloth_gemma-4-E4B-medical-o1-reasoning-SFT-merged",
+  "bias": "none",
+  "corda_config": null,
+  "ensure_weight_tying": false,
+  "eva_config": null,
+  "exclude_modules": null,
+  "fan_in_fan_out": false,
+  "inference_mode": true,
+  "init_lora_weights": true,
+  "layer_replication": null,
+  "layers_pattern": null,
+  "layers_to_transform": null,
+  "loftq_config": {},
+  "lora_alpha": 32,
+  "lora_bias": false,
+  "lora_dropout": 0.0,
+  "megatron_config": null,
+  "megatron_core": "megatron.core",
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "peft_version": "0.18.1",
+  "qalora_group_size": 16,
+  "r": 16,
+  "rank_pattern": {},
+  "revision": null,
+  "target_modules": [
+    "o_proj",
+    "down_proj",
+    "k_proj",
+    "up_proj",
+    "q_proj",
+    "v_proj",
+    "gate_proj"
+  ],
+  "target_parameters": null,
+  "task_type": "CAUSAL_LM",
+  "trainable_token_indices": null,
+  "use_dora": false,
+  "use_qalora": false,
+  "use_rslora": false
+}

checkpoint-400/adapter_model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d4e6cc8de31aef6cf529dbfeea5b628c4985ffbe1a8e0f99ace2eaf4c6731197
+size 169741912

checkpoint-400/chat_template.jinja

@@ -0,0 +1,344 @@
+{%- macro format_parameters(properties, required) -%}
+    {%- set standard_keys = ['description', 'type', 'properties', 'required', 'nullable'] -%}
+    {%- set ns = namespace(found_first=false) -%}
+    {%- for key, value in properties | dictsort -%}
+        {%- set add_comma = false -%}
+        {%- if key not in standard_keys -%}
+            {%- if ns.found_first %},{% endif -%}
+            {%- set ns.found_first = true -%}
+            {{ key }}:{
+            {%- if value['description'] -%}
+                description:<|"|>{{ value['description'] }}<|"|>
+                {%- set add_comma = true -%}
+            {%- endif -%}
+            {%- if value['type'] | upper == 'STRING' -%}
+                {%- if value['enum'] -%}
+                    {%- if add_comma %},{%- else -%} {%- set add_comma = true -%} {% endif -%}
+                    enum:{{ format_argument(value['enum']) }}
+                {%- endif -%}
+            {%- elif value['type'] | upper == 'ARRAY' -%}
+                {%- if value['items'] is mapping and value['items'] -%}
+                    {%- if add_comma %},{%- else -%} {%- set add_comma = true -%} {% endif -%}
+                    items:{
+                    {%- set ns_items = namespace(found_first=false) -%}
+                    {%- for item_key, item_value in value['items'] | dictsort -%}
+                        {%- if item_value is not none -%}
+                            {%- if ns_items.found_first %},{% endif -%}
+                            {%- set ns_items.found_first = true -%}
+                            {%- if item_key == 'properties' -%}
+                                properties:{
+                                {%- if item_value is mapping -%}
+                                    {{- format_parameters(item_value, value['items']['required'] | default([])) -}}
+                                {%- endif -%}
+                                }
+                            {%- elif item_key == 'required' -%}
+                                required:[
+                                {%- for req_item in item_value -%}
+                                    <|"|>{{- req_item -}}<|"|>
+                                    {%- if not loop.last %},{% endif -%}
+                                {%- endfor -%}
+                                ]
+                            {%- elif item_key == 'type' -%}
+                                {%- if item_value is string -%}
+                                    type:{{ format_argument(item_value | upper) }}
+                                {%- else -%}
+                                    type:{{ format_argument(item_value | map('upper') | list) }}
+                                {%- endif -%}
+                            {%- else -%}
+                                {{ item_key }}:{{ format_argument(item_value) }}
+                            {%- endif -%}
+                        {%- endif -%}
+                    {%- endfor -%}
+                    }
+                {%- endif -%}
+            {%- endif -%}
+            {%- if value['nullable'] %}
+                {%- if add_comma %},{%- else -%} {%- set add_comma = true -%} {% endif -%}
+                nullable:true
+            {%- endif -%}
+            {%- if value['type'] | upper == 'OBJECT' -%}
+                {%- if value['properties'] is defined and value['properties'] is mapping -%}
+                    {%- if add_comma %},{%- else -%} {%- set add_comma = true -%} {% endif -%}
+                    properties:{
+                    {{- format_parameters(value['properties'], value['required'] | default([])) -}}
+                    }
+                {%- elif value is mapping -%}
+                    {%- if add_comma %},{%- else -%} {%- set add_comma = true -%} {% endif -%}
+                    properties:{
+                    {{- format_parameters(value, value['required'] | default([])) -}}
+                    }
+                {%- endif -%}
+                {%- if value['required'] -%}
+                    {%- if add_comma %},{%- else -%} {%- set add_comma = true -%} {% endif -%}
+                    required:[
+                    {%- for item in value['required'] | default([]) -%}
+                        <|"|>{{- item -}}<|"|>
+                        {%- if not loop.last %},{% endif -%}
+                    {%- endfor -%}
+                    ]
+                {%- endif -%}
+            {%- endif -%}
+            {%- if add_comma %},{%- else -%} {%- set add_comma = true -%} {% endif -%}
+            type:<|"|>{{ value['type'] | upper }}<|"|>}
+        {%- endif -%}
+    {%- endfor -%}
+{%- endmacro -%}
+{%- macro format_function_declaration(tool_data) -%}
+    declaration:{{- tool_data['function']['name'] -}}{description:<|"|>{{- tool_data['function']['description'] -}}<|"|>
+    {%- set params = tool_data['function']['parameters'] -%}
+    {%- if params -%}
+        ,parameters:{
+        {%- if params['properties'] -%}
+            properties:{ {{- format_parameters(params['properties'], params['required']) -}} },
+        {%- endif -%}
+        {%- if params['required'] -%}
+            required:[
+            {%- for item in params['required'] -%}
+                <|"|>{{- item -}}<|"|>
+                {{- ',' if not loop.last -}}
+            {%- endfor -%}
+            ],
+        {%- endif -%}
+        {%- if params['type'] -%}
+            type:<|"|>{{- params['type'] | upper -}}<|"|>}
+        {%- endif -%}
+    {%- endif -%}
+    {%- if 'response' in tool_data['function'] -%}
+        {%- set response_declaration = tool_data['function']['response'] -%}
+        ,response:{
+        {%- if response_declaration['description'] -%}
+            description:<|"|>{{- response_declaration['description'] -}}<|"|>,
+        {%- endif -%}
+        {%- if response_declaration['type'] | upper == 'OBJECT' -%}
+            type:<|"|>{{- response_declaration['type'] | upper -}}<|"|>}
+        {%- endif -%}
+    {%- endif -%}
+    }
+{%- endmacro -%}
+{%- macro format_argument(argument, escape_keys=True) -%}
+    {%- if argument is string -%}
+        {{- '<|"|>' + argument + '<|"|>' -}}
+    {%- elif argument is boolean -%}
+        {{- 'true' if argument else 'false' -}}
+    {%- elif argument is mapping -%}
+        {{- '{' -}}
+        {%- set ns = namespace(found_first=false) -%}
+        {%- for key, value in argument | dictsort -%}
+            {%- if ns.found_first %},{% endif -%}
+            {%- set ns.found_first = true -%}
+            {%- if escape_keys -%}
+                {{- '<|"|>' + key + '<|"|>' -}}
+            {%- else -%}
+                {{- key -}}
+            {%- endif -%}
+            :{{- format_argument(value, escape_keys=escape_keys) -}}
+        {%- endfor -%}
+        {{- '}' -}}
+    {%- elif argument is sequence -%}
+        {{- '[' -}}
+        {%- for item in argument -%}
+            {{- format_argument(item, escape_keys=escape_keys) -}}
+            {%- if not loop.last %},{% endif -%}
+        {%- endfor -%}
+        {{- ']' -}}
+    {%- else -%}
+        {{- argument -}}
+    {%- endif -%}
+{%- endmacro -%}
+{%- macro strip_thinking(text) -%}
+    {%- set ns = namespace(result='') -%}
+    {%- for part in text.split('<channel|>') -%}
+        {%- if '<|channel>' in part -%}
+            {%- set ns.result = ns.result + part.split('<|channel>')[0] -%}
+        {%- else -%}
+            {%- set ns.result = ns.result + part -%}
+        {%- endif -%}
+    {%- endfor -%}
+    {{- ns.result | trim -}}
+{%- endmacro -%}
+
+{%- macro format_tool_response_block(tool_name, response) -%}
+    {{- '<|tool_response>' -}}
+    {%- if response is mapping -%}
+        {{- 'response:' + tool_name + '{' -}}
+        {%- for key, value in response | dictsort -%}
+            {{- key -}}:{{- format_argument(value, escape_keys=False) -}}
+            {%- if not loop.last %},{% endif -%}
+        {%- endfor -%}
+        {{- '}' -}}
+    {%- else -%}
+        {{- 'response:' + tool_name + '{value:' + format_argument(response, escape_keys=False) + '}' -}}
+    {%- endif -%}
+    {{- '<tool_response|>' -}}
+{%- endmacro -%}
+
+{%- set ns = namespace(prev_message_type=None) -%}
+{%- set loop_messages = messages -%}
+{{- bos_token -}}
+{#- Handle System/Tool Definitions Block -#}
+{%- if (enable_thinking is defined and enable_thinking) or tools or messages[0]['role'] in ['system', 'developer'] -%}
+    {{- '<|turn>system\n' -}}
+
+    {#- Inject Thinking token at the very top of the FIRST system turn -#}
+    {%- if enable_thinking is defined and enable_thinking -%}
+        {{- '<|think|>\n' -}}
+        {%- set ns.prev_message_type = 'think' -%}
+    {%- endif -%}
+
+    {%- if messages[0]['role'] in ['system', 'developer'] -%}
+        {{- messages[0]['content'] | trim -}}
+        {%- set loop_messages = messages[1:] -%}
+    {%- endif -%}
+
+    {%- if tools -%}
+        {%- for tool in tools %}
+            {{- '<|tool>' -}}
+            {{- format_function_declaration(tool) | trim -}}
+            {{- '<tool|>' -}}
+        {%- endfor %}
+        {%- set ns.prev_message_type = 'tool' -%}
+    {%- endif -%}
+
+    {{- '<turn|>\n' -}}
+{%- endif %}
+
+{#- Pre-scan: find last user message index for reasoning guard -#}
+{%- set ns_turn = namespace(last_user_idx=-1) -%}
+{%- for i in range(loop_messages | length) -%}
+    {%- if loop_messages[i]['role'] == 'user' -%}
+        {%- set ns_turn.last_user_idx = i -%}
+    {%- endif -%}
+{%- endfor -%}
+
+{#- Loop through messages -#}
+{%- for message in loop_messages -%}
+    {%- if message['role'] != 'tool' -%}
+    {%- set ns.prev_message_type = None -%}
+    {%- set role = 'model' if message['role'] == 'assistant' else message['role'] -%}
+    {#- Detect continuation: suppress duplicate <|turn>model when previous non-tool message was also assistant -#}
+    {%- set prev_nt = namespace(role=None, found=false) -%}
+    {%- if loop.index0 > 0 -%}
+        {%- for j in range(loop.index0 - 1, -1, -1) -%}
+            {%- if not prev_nt.found -%}
+                {%- if loop_messages[j]['role'] != 'tool' -%}
+                    {%- set prev_nt.role = loop_messages[j]['role'] -%}
+                    {%- set prev_nt.found = true -%}
+                {%- endif -%}
+            {%- endif -%}
+        {%- endfor -%}
+    {%- endif -%}
+    {%- set continue_same_model_turn = (role == 'model' and prev_nt.role == 'assistant') -%}
+    {%- if not continue_same_model_turn -%}
+        {{- '<|turn>' + role + '\n' }}
+    {%- endif -%}
+
+    {#- Render reasoning/reasoning_content as thinking channel -#}
+    {%- set thinking_text = message.get('reasoning') or message.get('reasoning_content') -%}
+    {%- if thinking_text and loop.index0 > ns_turn.last_user_idx and message.get('tool_calls') -%}
+        {{- '<|channel>thought\n' + thinking_text + '\n<channel|>' -}}
+    {%- endif -%}
+
+            {%- if message['tool_calls'] -%}
+                {%- for tool_call in message['tool_calls'] -%}
+                    {%- set function = tool_call['function'] -%}
+                    {{- '<|tool_call>call:' + function['name'] + '{' -}}
+                    {%- if function['arguments'] is mapping -%}
+                        {%- set ns_args = namespace(found_first=false) -%}
+                        {%- for key, value in function['arguments'] | dictsort -%}
+                            {%- if ns_args.found_first %},{% endif -%}
+                            {%- set ns_args.found_first = true -%}
+                            {{- key -}}:{{- format_argument(value, escape_keys=False) -}}
+                        {%- endfor -%}
+                    {%- elif function['arguments'] is string -%}
+                        {{- function['arguments'] -}}
+                    {%- endif -%}
+                    {{- '}<tool_call|>' -}}
+                {%- endfor -%}
+                {%- set ns.prev_message_type = 'tool_call' -%}
+            {%- endif -%}
+
+            {%- set ns_tr_out = namespace(flag=false) -%}
+            {%- if message.get('tool_responses') -%}
+                {#- Legacy: tool_responses embedded on the assistant message (Google/Gemma native) -#}
+                {%- for tool_response in message['tool_responses'] -%}
+                    {{- format_tool_response_block(tool_response['name'] | default('unknown'), tool_response['response']) -}}
+                    {%- set ns_tr_out.flag = true -%}
+                    {%- set ns.prev_message_type = 'tool_response' -%}
+                {%- endfor -%}
+            {%- elif message.get('tool_calls') -%}
+                {#- OpenAI Chat Completions: forward-scan consecutive role:tool messages -#}
+                {%- set ns_tool_scan = namespace(stopped=false) -%}
+                {%- for k in range(loop.index0 + 1, loop_messages | length) -%}
+                    {%- if ns_tool_scan.stopped -%}
+                    {%- elif loop_messages[k]['role'] != 'tool' -%}
+                        {%- set ns_tool_scan.stopped = true -%}
+                    {%- else -%}
+                        {%- set follow = loop_messages[k] -%}
+                        {#- Resolve tool_call_id to function name -#}
+                        {%- set ns_tname = namespace(name=follow.get('name') | default('unknown')) -%}
+                        {%- for tc in message['tool_calls'] -%}
+                            {%- if tc.get('id') == follow.get('tool_call_id') -%}
+                                {%- set ns_tname.name = tc['function']['name'] -%}
+                            {%- endif -%}
+                        {%- endfor -%}
+                        {#- Handle content as string or content-parts array -#}
+                        {%- set tool_body = follow.get('content') -%}
+                        {%- if tool_body is string -%}
+                            {{- format_tool_response_block(ns_tname.name, tool_body) -}}
+                        {%- elif tool_body is sequence and tool_body is not string -%}
+                            {%- set ns_txt = namespace(s='') -%}
+                            {%- for part in tool_body -%}
+                                {%- if part.get('type') == 'text' -%}
+                                    {%- set ns_txt.s = ns_txt.s + (part.get('text') | default('')) -%}
+                                {%- endif -%}
+                            {%- endfor -%}
+                            {{- format_tool_response_block(ns_tname.name, ns_txt.s) -}}
+                        {%- else -%}
+                            {{- format_tool_response_block(ns_tname.name, tool_body) -}}
+                        {%- endif -%}
+                        {%- set ns_tr_out.flag = true -%}
+                        {%- set ns.prev_message_type = 'tool_response' -%}
+                    {%- endif -%}
+                {%- endfor -%}
+            {%- endif -%}
+
+            {%- if message['content'] is string -%}
+                {%- if role == 'model' -%}
+                    {{- strip_thinking(message['content']) -}}
+                {%- else -%}
+                    {{- message['content'] | trim -}}
+                {%- endif -%}
+            {%- elif message['content'] is sequence -%}
+                {%- for item in message['content'] -%}
+                    {%- if item['type'] == 'text' -%}
+                        {%- if role == 'model' -%}
+                            {{- strip_thinking(item['text']) -}}
+                        {%- else -%}
+                            {{- item['text'] | trim -}}
+                        {%- endif -%}
+                    {%- elif item['type'] == 'image' -%}
+                        {{- '<|image|>' -}}
+                        {%- set ns.prev_message_type = 'image' -%}
+                    {%- elif item['type'] == 'audio' -%}
+                        {{- '<|audio|>' -}}
+                        {%- set ns.prev_message_type = 'audio' -%}
+                    {%- elif item['type'] == 'video' -%}
+                        {{- '<|video|>' -}}
+                        {%- set ns.prev_message_type = 'video' -%}
+                    {%- endif -%}
+                {%- endfor -%}
+            {%- endif -%}
+
+        {%- if ns.prev_message_type == 'tool_call' and not ns_tr_out.flag -%}
+            {{- '<|tool_response>' -}}
+        {%- elif not (ns_tr_out.flag and not message.get('content')) -%}
+            {{- '<turn|>\n' -}}
+        {%- endif -%}
+    {%- endif -%}
+{%- endfor -%}
+
+{%- if add_generation_prompt -%}
+    {%- if ns.prev_message_type != 'tool_response' and ns.prev_message_type != 'tool_call' -%}
+        {{- '<|turn>model\n' -}}
+    {%- endif -%}
+{%- endif -%}

checkpoint-400/optimizer.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2eeb3920550c1da4c0d2d4daddc085b8608d13069ab11b9ae2ced297a8965578
+size 75237013

checkpoint-400/processor_config.json

@@ -0,0 +1,75 @@
+{
+  "audio_ms_per_token": 40,
+  "audio_seq_length": 750,
+  "feature_extractor": {
+    "dither": 0.0,
+    "feature_extractor_type": "Gemma4AudioFeatureExtractor",
+    "feature_size": 128,
+    "fft_length": 512,
+    "fft_overdrive": false,
+    "frame_length": 320,
+    "hop_length": 160,
+    "input_scale_factor": 1.0,
+    "max_frequency": 8000.0,
+    "mel_floor": 0.001,
+    "min_frequency": 0.0,
+    "padding_side": "left",
+    "padding_value": 0.0,
+    "per_bin_mean": null,
+    "per_bin_stddev": null,
+    "preemphasis": 0.0,
+    "preemphasis_htk_flavor": true,
+    "return_attention_mask": true,
+    "sampling_rate": 16000
+  },
+  "image_processor": {
+    "do_convert_rgb": true,
+    "do_normalize": false,
+    "do_rescale": true,
+    "do_resize": true,
+    "image_mean": [
+      0.0,
+      0.0,
+      0.0
+    ],
+    "image_processor_type": "Gemma4ImageProcessor",
+    "image_seq_length": 280,
+    "image_std": [
+      1.0,
+      1.0,
+      1.0
+    ],
+    "max_soft_tokens": 280,
+    "patch_size": 16,
+    "pooling_kernel_size": 3,
+    "resample": 3,
+    "rescale_factor": 0.00392156862745098
+  },
+  "image_seq_length": 280,
+  "processor_class": "Gemma4Processor",
+  "video_processor": {
+    "do_convert_rgb": true,
+    "do_normalize": true,
+    "do_rescale": true,
+    "do_resize": true,
+    "do_sample_frames": true,
+    "image_mean": [
+      0.0,
+      0.0,
+      0.0
+    ],
+    "image_std": [
+      1.0,
+      1.0,
+      1.0
+    ],
+    "max_soft_tokens": 70,
+    "num_frames": 32,
+    "patch_size": 16,
+    "pooling_kernel_size": 3,
+    "resample": 3,
+    "rescale_factor": 0.00392156862745098,
+    "return_metadata": false,
+    "video_processor_type": "Gemma4VideoProcessor"
+  }
+}

checkpoint-400/rng_state.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7c800b778fa7e115e4c34de8529902de8b61c9a1b4bab3eb8295d06dafff030e
+size 14645

checkpoint-400/scheduler.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4965495586d9408c759f5ffce6e70e6fea6ebdb669bf96754a7945e45f7fc8e8
+size 1465

checkpoint-400/tokenizer.json

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:cc8d3a0ce36466ccc1278bf987df5f71db1719b9ca6b4118264f45cb627bfe0f
+size 32169626

checkpoint-400/tokenizer_config.json

@@ -0,0 +1,95 @@
+{
+  "audio_token": "<|audio|>",
+  "backend": "tokenizers",
+  "boa_token": "<|audio>",
+  "boi_token": "<|image>",
+  "bos_token": "<bos>",
+  "eoa_token": "<audio|>",
+  "eoc_token": "<channel|>",
+  "eoi_token": "<image|>",
+  "eos_token": "<turn|>",
+  "eot_token": "<turn|>",
+  "escape_token": "<|\"|>",
+  "etc_token": "<tool_call|>",
+  "etd_token": "<tool|>",
+  "etr_token": "<tool_response|>",
+  "extra_special_tokens": [
+    "<|video|>"
+  ],
+  "image_token": "<|image|>",
+  "is_local": false,
+  "mask_token": "<mask>",
+  "model_max_length": 131072,
+  "model_specific_special_tokens": {
+    "audio_token": "<|audio|>",
+    "boa_token": "<|audio>",
+    "boi_token": "<|image>",
+    "eoa_token": "<audio|>",
+    "eoc_token": "<channel|>",
+    "eoi_token": "<image|>",
+    "eot_token": "<turn|>",
+    "escape_token": "<|\"|>",
+    "etc_token": "<tool_call|>",
+    "etd_token": "<tool|>",
+    "etr_token": "<tool_response|>",
+    "image_token": "<|image|>",
+    "soc_token": "<|channel>",
+    "sot_token": "<|turn>",
+    "stc_token": "<|tool_call>",
+    "std_token": "<|tool>",
+    "str_token": "<|tool_response>",
+    "think_token": "<|think|>"
+  },
+  "pad_token": "<pad>",
+  "padding_side": "right",
+  "processor_class": "Gemma4Processor",
+  "response_schema": {
+    "properties": {
+      "content": {
+        "type": "string"
+      },
+      "role": {
+        "const": "assistant"
+      },
+      "thinking": {
+        "type": "string"
+      },
+      "tool_calls": {
+        "items": {
+          "properties": {
+            "function": {
+              "properties": {
+                "arguments": {
+                  "additionalProperties": {},
+                  "type": "object",
+                  "x-parser": "gemma4-tool-call"
+                },
+                "name": {
+                  "type": "string"
+                }
+              },
+              "type": "object",
+              "x-regex": "call\\:(?P<name>\\w+)(?P<arguments>\\{.*\\})"
+            },
+            "type": {
+              "const": "function"
+            }
+          },
+          "type": "object"
+        },
+        "type": "array",
+        "x-regex-iterator": "<\\|tool_call>(.*?)<tool_call\\|>"
+      }
+    },
+    "type": "object",
+    "x-regex": "(\\<\\|channel\\>thought\\n(?P<thinking>.*?)\\<channel\\|\\>)?(?P<tool_calls>\\<\\|tool_call\\>.*\\<tool_call\\|\\>)?(?P<content>(?:(?!\\<turn\\|\\>)(?!\\<\\|tool_response\\>).)+)?(?:\\<turn\\|\\>|\\<\\|tool_response\\>)?"
+  },
+  "soc_token": "<|channel>",
+  "sot_token": "<|turn>",
+  "stc_token": "<|tool_call>",
+  "std_token": "<|tool>",
+  "str_token": "<|tool_response>",
+  "think_token": "<|think|>",
+  "tokenizer_class": "GemmaTokenizer",
+  "unk_token": "<unk>"
+}

checkpoint-400/training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ae7a694dd4875ee97139bdd8ebc7349b14ce4d529c119642d14f7f7f5ebd9df4
+size 5777

checkpoint-600/README.md

@@ -0,0 +1,210 @@
+---
+base_model: chenhaodev/unsloth_gemma-4-E4B-medical-o1-reasoning-SFT-merged
+library_name: peft
+pipeline_tag: text-generation
+tags:
+- base_model:adapter:chenhaodev/unsloth_gemma-4-E4B-medical-o1-reasoning-SFT-merged
+- lora
+- sft
+- transformers
+- trl
+- unsloth
+---
+
+# Model Card for Model ID
+
+<!-- Provide a quick summary of what the model is/does. -->
+
+
+
+## Model Details
+
+### Model Description
+
+<!-- Provide a longer summary of what this model is. -->
+
+
+
+- **Developed by:** [More Information Needed]
+- **Funded by [optional]:** [More Information Needed]
+- **Shared by [optional]:** [More Information Needed]
+- **Model type:** [More Information Needed]
+- **Language(s) (NLP):** [More Information Needed]
+- **License:** [More Information Needed]
+- **Finetuned from model [optional]:** [More Information Needed]
+
+### Model Sources [optional]
+
+<!-- Provide the basic links for the model. -->
+
+- **Repository:** [More Information Needed]
+- **Paper [optional]:** [More Information Needed]
+- **Demo [optional]:** [More Information Needed]
+
+## Uses
+
+<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
+
+### Direct Use
+
+<!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
+
+[More Information Needed]
+
+### Downstream Use [optional]
+
+<!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->
+
+[More Information Needed]
+
+### Out-of-Scope Use
+
+<!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
+
+[More Information Needed]
+
+## Bias, Risks, and Limitations
+
+<!-- This section is meant to convey both technical and sociotechnical limitations. -->
+
+[More Information Needed]
+
+### Recommendations
+
+<!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->
+
+Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
+
+## How to Get Started with the Model
+
+Use the code below to get started with the model.
+
+[More Information Needed]
+
+## Training Details
+
+### Training Data
+
+<!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
+
+[More Information Needed]
+
+### Training Procedure
+
+<!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
+
+#### Preprocessing [optional]
+
+[More Information Needed]
+
+
+#### Training Hyperparameters
+
+- **Training regime:** [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->
+
+#### Speeds, Sizes, Times [optional]
+
+<!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
+
+[More Information Needed]
+
+## Evaluation
+
+<!-- This section describes the evaluation protocols and provides the results. -->
+
+### Testing Data, Factors & Metrics
+
+#### Testing Data
+
+<!-- This should link to a Dataset Card if possible. -->
+
+[More Information Needed]
+
+#### Factors
+
+<!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->
+
+[More Information Needed]
+
+#### Metrics
+
+<!-- These are the evaluation metrics being used, ideally with a description of why. -->
+
+[More Information Needed]
+
+### Results
+
+[More Information Needed]
+
+#### Summary
+
+
+
+## Model Examination [optional]
+
+<!-- Relevant interpretability work for the model goes here -->
+
+[More Information Needed]
+
+## Environmental Impact
+
+<!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly -->
+
+Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700).
+
+- **Hardware Type:** [More Information Needed]
+- **Hours used:** [More Information Needed]
+- **Cloud Provider:** [More Information Needed]
+- **Compute Region:** [More Information Needed]
+- **Carbon Emitted:** [More Information Needed]
+
+## Technical Specifications [optional]
+
+### Model Architecture and Objective
+
+[More Information Needed]
+
+### Compute Infrastructure
+
+[More Information Needed]
+
+#### Hardware
+
+[More Information Needed]
+
+#### Software
+
+[More Information Needed]
+
+## Citation [optional]
+
+<!-- If there is a paper or blog post introducing the model, the APA and Bibtex information for that should go in this section. -->
+
+**BibTeX:**
+
+[More Information Needed]
+
+**APA:**
+
+[More Information Needed]
+
+## Glossary [optional]
+
+<!-- If relevant, include terms and calculations in this section that can help readers understand the model or model card. -->
+
+[More Information Needed]
+
+## More Information [optional]
+
+[More Information Needed]
+
+## Model Card Authors [optional]
+
+[More Information Needed]
+
+## Model Card Contact
+
+[More Information Needed]
+### Framework versions
+
+- PEFT 0.18.1

checkpoint-600/adapter_config.json

@@ -0,0 +1,50 @@
+{
+  "alora_invocation_tokens": null,
+  "alpha_pattern": {},
+  "arrow_config": null,
+  "auto_mapping": {
+    "base_model_class": "Gemma4ForConditionalGeneration",
+    "parent_library": "transformers.models.gemma4.modeling_gemma4",
+    "unsloth_fixed": true
+  },
+  "base_model_name_or_path": "chenhaodev/unsloth_gemma-4-E4B-medical-o1-reasoning-SFT-merged",
+  "bias": "none",
+  "corda_config": null,
+  "ensure_weight_tying": false,
+  "eva_config": null,
+  "exclude_modules": null,
+  "fan_in_fan_out": false,
+  "inference_mode": true,
+  "init_lora_weights": true,
+  "layer_replication": null,
+  "layers_pattern": null,
+  "layers_to_transform": null,
+  "loftq_config": {},
+  "lora_alpha": 32,
+  "lora_bias": false,
+  "lora_dropout": 0.0,
+  "megatron_config": null,
+  "megatron_core": "megatron.core",
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "peft_version": "0.18.1",
+  "qalora_group_size": 16,
+  "r": 16,
+  "rank_pattern": {},
+  "revision": null,
+  "target_modules": [
+    "o_proj",
+    "down_proj",
+    "k_proj",
+    "up_proj",
+    "q_proj",
+    "v_proj",
+    "gate_proj"
+  ],
+  "target_parameters": null,
+  "task_type": "CAUSAL_LM",
+  "trainable_token_indices": null,
+  "use_dora": false,
+  "use_qalora": false,
+  "use_rslora": false
+}

checkpoint-600/adapter_model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:aa0e5e8019c6a4ef49eb95d35d95a27bc58f0b4877ef5182f7977bb3eb313f68
+size 169741912

checkpoint-600/chat_template.jinja

@@ -0,0 +1,344 @@
+{%- macro format_parameters(properties, required) -%}
+    {%- set standard_keys = ['description', 'type', 'properties', 'required', 'nullable'] -%}
+    {%- set ns = namespace(found_first=false) -%}
+    {%- for key, value in properties | dictsort -%}
+        {%- set add_comma = false -%}
+        {%- if key not in standard_keys -%}
+            {%- if ns.found_first %},{% endif -%}
+            {%- set ns.found_first = true -%}
+            {{ key }}:{
+            {%- if value['description'] -%}
+                description:<|"|>{{ value['description'] }}<|"|>
+                {%- set add_comma = true -%}
+            {%- endif -%}
+            {%- if value['type'] | upper == 'STRING' -%}
+                {%- if value['enum'] -%}
+                    {%- if add_comma %},{%- else -%} {%- set add_comma = true -%} {% endif -%}
+                    enum:{{ format_argument(value['enum']) }}
+                {%- endif -%}
+            {%- elif value['type'] | upper == 'ARRAY' -%}
+                {%- if value['items'] is mapping and value['items'] -%}
+                    {%- if add_comma %},{%- else -%} {%- set add_comma = true -%} {% endif -%}
+                    items:{
+                    {%- set ns_items = namespace(found_first=false) -%}
+                    {%- for item_key, item_value in value['items'] | dictsort -%}
+                        {%- if item_value is not none -%}
+                            {%- if ns_items.found_first %},{% endif -%}
+                            {%- set ns_items.found_first = true -%}
+                            {%- if item_key == 'properties' -%}
+                                properties:{
+                                {%- if item_value is mapping -%}
+                                    {{- format_parameters(item_value, value['items']['required'] | default([])) -}}
+                                {%- endif -%}
+                                }
+                            {%- elif item_key == 'required' -%}
+                                required:[
+                                {%- for req_item in item_value -%}
+                                    <|"|>{{- req_item -}}<|"|>
+                                    {%- if not loop.last %},{% endif -%}
+                                {%- endfor -%}
+                                ]
+                            {%- elif item_key == 'type' -%}
+                                {%- if item_value is string -%}
+                                    type:{{ format_argument(item_value | upper) }}
+                                {%- else -%}
+                                    type:{{ format_argument(item_value | map('upper') | list) }}
+                                {%- endif -%}
+                            {%- else -%}
+                                {{ item_key }}:{{ format_argument(item_value) }}
+                            {%- endif -%}
+                        {%- endif -%}
+                    {%- endfor -%}
+                    }
+                {%- endif -%}
+            {%- endif -%}
+            {%- if value['nullable'] %}
+                {%- if add_comma %},{%- else -%} {%- set add_comma = true -%} {% endif -%}
+                nullable:true
+            {%- endif -%}
+            {%- if value['type'] | upper == 'OBJECT' -%}
+                {%- if value['properties'] is defined and value['properties'] is mapping -%}
+                    {%- if add_comma %},{%- else -%} {%- set add_comma = true -%} {% endif -%}
+                    properties:{
+                    {{- format_parameters(value['properties'], value['required'] | default([])) -}}
+                    }
+                {%- elif value is mapping -%}
+                    {%- if add_comma %},{%- else -%} {%- set add_comma = true -%} {% endif -%}
+                    properties:{
+                    {{- format_parameters(value, value['required'] | default([])) -}}
+                    }
+                {%- endif -%}
+                {%- if value['required'] -%}
+                    {%- if add_comma %},{%- else -%} {%- set add_comma = true -%} {% endif -%}
+                    required:[
+                    {%- for item in value['required'] | default([]) -%}
+                        <|"|>{{- item -}}<|"|>
+                        {%- if not loop.last %},{% endif -%}
+                    {%- endfor -%}
+                    ]
+                {%- endif -%}
+            {%- endif -%}
+            {%- if add_comma %},{%- else -%} {%- set add_comma = true -%} {% endif -%}
+            type:<|"|>{{ value['type'] | upper }}<|"|>}
+        {%- endif -%}
+    {%- endfor -%}
+{%- endmacro -%}
+{%- macro format_function_declaration(tool_data) -%}
+    declaration:{{- tool_data['function']['name'] -}}{description:<|"|>{{- tool_data['function']['description'] -}}<|"|>
+    {%- set params = tool_data['function']['parameters'] -%}
+    {%- if params -%}
+        ,parameters:{
+        {%- if params['properties'] -%}
+            properties:{ {{- format_parameters(params['properties'], params['required']) -}} },
+        {%- endif -%}
+        {%- if params['required'] -%}
+            required:[
+            {%- for item in params['required'] -%}
+                <|"|>{{- item -}}<|"|>
+                {{- ',' if not loop.last -}}
+            {%- endfor -%}
+            ],
+        {%- endif -%}
+        {%- if params['type'] -%}
+            type:<|"|>{{- params['type'] | upper -}}<|"|>}
+        {%- endif -%}
+    {%- endif -%}
+    {%- if 'response' in tool_data['function'] -%}
+        {%- set response_declaration = tool_data['function']['response'] -%}
+        ,response:{
+        {%- if response_declaration['description'] -%}
+            description:<|"|>{{- response_declaration['description'] -}}<|"|>,
+        {%- endif -%}
+        {%- if response_declaration['type'] | upper == 'OBJECT' -%}
+            type:<|"|>{{- response_declaration['type'] | upper -}}<|"|>}
+        {%- endif -%}
+    {%- endif -%}
+    }
+{%- endmacro -%}
+{%- macro format_argument(argument, escape_keys=True) -%}
+    {%- if argument is string -%}
+        {{- '<|"|>' + argument + '<|"|>' -}}
+    {%- elif argument is boolean -%}
+        {{- 'true' if argument else 'false' -}}
+    {%- elif argument is mapping -%}
+        {{- '{' -}}
+        {%- set ns = namespace(found_first=false) -%}
+        {%- for key, value in argument | dictsort -%}
+            {%- if ns.found_first %},{% endif -%}
+            {%- set ns.found_first = true -%}
+            {%- if escape_keys -%}
+                {{- '<|"|>' + key + '<|"|>' -}}
+            {%- else -%}
+                {{- key -}}
+            {%- endif -%}
+            :{{- format_argument(value, escape_keys=escape_keys) -}}
+        {%- endfor -%}
+        {{- '}' -}}
+    {%- elif argument is sequence -%}
+        {{- '[' -}}
+        {%- for item in argument -%}
+            {{- format_argument(item, escape_keys=escape_keys) -}}
+            {%- if not loop.last %},{% endif -%}
+        {%- endfor -%}
+        {{- ']' -}}
+    {%- else -%}
+        {{- argument -}}
+    {%- endif -%}
+{%- endmacro -%}
+{%- macro strip_thinking(text) -%}
+    {%- set ns = namespace(result='') -%}
+    {%- for part in text.split('<channel|>') -%}
+        {%- if '<|channel>' in part -%}
+            {%- set ns.result = ns.result + part.split('<|channel>')[0] -%}
+        {%- else -%}
+            {%- set ns.result = ns.result + part -%}
+        {%- endif -%}
+    {%- endfor -%}
+    {{- ns.result | trim -}}
+{%- endmacro -%}
+
+{%- macro format_tool_response_block(tool_name, response) -%}
+    {{- '<|tool_response>' -}}
+    {%- if response is mapping -%}
+        {{- 'response:' + tool_name + '{' -}}
+        {%- for key, value in response | dictsort -%}
+            {{- key -}}:{{- format_argument(value, escape_keys=False) -}}
+            {%- if not loop.last %},{% endif -%}
+        {%- endfor -%}
+        {{- '}' -}}
+    {%- else -%}
+        {{- 'response:' + tool_name + '{value:' + format_argument(response, escape_keys=False) + '}' -}}
+    {%- endif -%}
+    {{- '<tool_response|>' -}}
+{%- endmacro -%}
+
+{%- set ns = namespace(prev_message_type=None) -%}
+{%- set loop_messages = messages -%}
+{{- bos_token -}}
+{#- Handle System/Tool Definitions Block -#}
+{%- if (enable_thinking is defined and enable_thinking) or tools or messages[0]['role'] in ['system', 'developer'] -%}
+    {{- '<|turn>system\n' -}}
+
+    {#- Inject Thinking token at the very top of the FIRST system turn -#}
+    {%- if enable_thinking is defined and enable_thinking -%}
+        {{- '<|think|>\n' -}}
+        {%- set ns.prev_message_type = 'think' -%}
+    {%- endif -%}
+
+    {%- if messages[0]['role'] in ['system', 'developer'] -%}
+        {{- messages[0]['content'] | trim -}}
+        {%- set loop_messages = messages[1:] -%}
+    {%- endif -%}
+
+    {%- if tools -%}
+        {%- for tool in tools %}
+            {{- '<|tool>' -}}
+            {{- format_function_declaration(tool) | trim -}}
+            {{- '<tool|>' -}}
+        {%- endfor %}
+        {%- set ns.prev_message_type = 'tool' -%}
+    {%- endif -%}
+
+    {{- '<turn|>\n' -}}
+{%- endif %}
+
+{#- Pre-scan: find last user message index for reasoning guard -#}
+{%- set ns_turn = namespace(last_user_idx=-1) -%}
+{%- for i in range(loop_messages | length) -%}
+    {%- if loop_messages[i]['role'] == 'user' -%}
+        {%- set ns_turn.last_user_idx = i -%}
+    {%- endif -%}
+{%- endfor -%}
+
+{#- Loop through messages -#}
+{%- for message in loop_messages -%}
+    {%- if message['role'] != 'tool' -%}
+    {%- set ns.prev_message_type = None -%}
+    {%- set role = 'model' if message['role'] == 'assistant' else message['role'] -%}
+    {#- Detect continuation: suppress duplicate <|turn>model when previous non-tool message was also assistant -#}
+    {%- set prev_nt = namespace(role=None, found=false) -%}
+    {%- if loop.index0 > 0 -%}
+        {%- for j in range(loop.index0 - 1, -1, -1) -%}
+            {%- if not prev_nt.found -%}
+                {%- if loop_messages[j]['role'] != 'tool' -%}
+                    {%- set prev_nt.role = loop_messages[j]['role'] -%}
+                    {%- set prev_nt.found = true -%}
+                {%- endif -%}
+            {%- endif -%}
+        {%- endfor -%}
+    {%- endif -%}
+    {%- set continue_same_model_turn = (role == 'model' and prev_nt.role == 'assistant') -%}
+    {%- if not continue_same_model_turn -%}
+        {{- '<|turn>' + role + '\n' }}
+    {%- endif -%}
+
+    {#- Render reasoning/reasoning_content as thinking channel -#}
+    {%- set thinking_text = message.get('reasoning') or message.get('reasoning_content') -%}
+    {%- if thinking_text and loop.index0 > ns_turn.last_user_idx and message.get('tool_calls') -%}
+        {{- '<|channel>thought\n' + thinking_text + '\n<channel|>' -}}
+    {%- endif -%}
+
+            {%- if message['tool_calls'] -%}
+                {%- for tool_call in message['tool_calls'] -%}
+                    {%- set function = tool_call['function'] -%}
+                    {{- '<|tool_call>call:' + function['name'] + '{' -}}
+                    {%- if function['arguments'] is mapping -%}
+                        {%- set ns_args = namespace(found_first=false) -%}
+                        {%- for key, value in function['arguments'] | dictsort -%}
+                            {%- if ns_args.found_first %},{% endif -%}
+                            {%- set ns_args.found_first = true -%}
+                            {{- key -}}:{{- format_argument(value, escape_keys=False) -}}
+                        {%- endfor -%}
+                    {%- elif function['arguments'] is string -%}
+                        {{- function['arguments'] -}}
+                    {%- endif -%}
+                    {{- '}<tool_call|>' -}}
+                {%- endfor -%}
+                {%- set ns.prev_message_type = 'tool_call' -%}
+            {%- endif -%}
+
+            {%- set ns_tr_out = namespace(flag=false) -%}
+            {%- if message.get('tool_responses') -%}
+                {#- Legacy: tool_responses embedded on the assistant message (Google/Gemma native) -#}
+                {%- for tool_response in message['tool_responses'] -%}
+                    {{- format_tool_response_block(tool_response['name'] | default('unknown'), tool_response['response']) -}}
+                    {%- set ns_tr_out.flag = true -%}
+                    {%- set ns.prev_message_type = 'tool_response' -%}
+                {%- endfor -%}
+            {%- elif message.get('tool_calls') -%}
+                {#- OpenAI Chat Completions: forward-scan consecutive role:tool messages -#}
+                {%- set ns_tool_scan = namespace(stopped=false) -%}
+                {%- for k in range(loop.index0 + 1, loop_messages | length) -%}
+                    {%- if ns_tool_scan.stopped -%}
+                    {%- elif loop_messages[k]['role'] != 'tool' -%}
+                        {%- set ns_tool_scan.stopped = true -%}
+                    {%- else -%}
+                        {%- set follow = loop_messages[k] -%}
+                        {#- Resolve tool_call_id to function name -#}
+                        {%- set ns_tname = namespace(name=follow.get('name') | default('unknown')) -%}
+                        {%- for tc in message['tool_calls'] -%}
+                            {%- if tc.get('id') == follow.get('tool_call_id') -%}
+                                {%- set ns_tname.name = tc['function']['name'] -%}
+                            {%- endif -%}
+                        {%- endfor -%}
+                        {#- Handle content as string or content-parts array -#}
+                        {%- set tool_body = follow.get('content') -%}
+                        {%- if tool_body is string -%}
+                            {{- format_tool_response_block(ns_tname.name, tool_body) -}}
+                        {%- elif tool_body is sequence and tool_body is not string -%}
+                            {%- set ns_txt = namespace(s='') -%}
+                            {%- for part in tool_body -%}
+                                {%- if part.get('type') == 'text' -%}
+                                    {%- set ns_txt.s = ns_txt.s + (part.get('text') | default('')) -%}
+                                {%- endif -%}
+                            {%- endfor -%}
+                            {{- format_tool_response_block(ns_tname.name, ns_txt.s) -}}
+                        {%- else -%}
+                            {{- format_tool_response_block(ns_tname.name, tool_body) -}}
+                        {%- endif -%}
+                        {%- set ns_tr_out.flag = true -%}
+                        {%- set ns.prev_message_type = 'tool_response' -%}
+                    {%- endif -%}
+                {%- endfor -%}
+            {%- endif -%}
+
+            {%- if message['content'] is string -%}
+                {%- if role == 'model' -%}
+                    {{- strip_thinking(message['content']) -}}
+                {%- else -%}
+                    {{- message['content'] | trim -}}
+                {%- endif -%}
+            {%- elif message['content'] is sequence -%}
+                {%- for item in message['content'] -%}
+                    {%- if item['type'] == 'text' -%}
+                        {%- if role == 'model' -%}
+                            {{- strip_thinking(item['text']) -}}
+                        {%- else -%}
+                            {{- item['text'] | trim -}}
+                        {%- endif -%}
+                    {%- elif item['type'] == 'image' -%}
+                        {{- '<|image|>' -}}
+                        {%- set ns.prev_message_type = 'image' -%}
+                    {%- elif item['type'] == 'audio' -%}
+                        {{- '<|audio|>' -}}
+                        {%- set ns.prev_message_type = 'audio' -%}
+                    {%- elif item['type'] == 'video' -%}
+                        {{- '<|video|>' -}}
+                        {%- set ns.prev_message_type = 'video' -%}
+                    {%- endif -%}
+                {%- endfor -%}
+            {%- endif -%}
+
+        {%- if ns.prev_message_type == 'tool_call' and not ns_tr_out.flag -%}
+            {{- '<|tool_response>' -}}
+        {%- elif not (ns_tr_out.flag and not message.get('content')) -%}
+            {{- '<turn|>\n' -}}
+        {%- endif -%}
+    {%- endif -%}
+{%- endfor -%}
+
+{%- if add_generation_prompt -%}
+    {%- if ns.prev_message_type != 'tool_response' and ns.prev_message_type != 'tool_call' -%}
+        {{- '<|turn>model\n' -}}
+    {%- endif -%}
+{%- endif -%}

checkpoint-600/optimizer.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4b4e9be09f94098841eb75087583d9a8d207b22f4d40d49fd88693fea1ab1298
+size 75237013

checkpoint-600/processor_config.json

@@ -0,0 +1,75 @@
+{
+  "audio_ms_per_token": 40,
+  "audio_seq_length": 750,
+  "feature_extractor": {
+    "dither": 0.0,
+    "feature_extractor_type": "Gemma4AudioFeatureExtractor",
+    "feature_size": 128,
+    "fft_length": 512,
+    "fft_overdrive": false,
+    "frame_length": 320,
+    "hop_length": 160,
+    "input_scale_factor": 1.0,
+    "max_frequency": 8000.0,
+    "mel_floor": 0.001,
+    "min_frequency": 0.0,
+    "padding_side": "left",
+    "padding_value": 0.0,
+    "per_bin_mean": null,
+    "per_bin_stddev": null,
+    "preemphasis": 0.0,
+    "preemphasis_htk_flavor": true,
+    "return_attention_mask": true,
+    "sampling_rate": 16000
+  },
+  "image_processor": {
+    "do_convert_rgb": true,
+    "do_normalize": false,
+    "do_rescale": true,
+    "do_resize": true,
+    "image_mean": [
+      0.0,
+      0.0,
+      0.0
+    ],
+    "image_processor_type": "Gemma4ImageProcessor",
+    "image_seq_length": 280,
+    "image_std": [
+      1.0,
+      1.0,
+      1.0
+    ],
+    "max_soft_tokens": 280,
+    "patch_size": 16,
+    "pooling_kernel_size": 3,
+    "resample": 3,
+    "rescale_factor": 0.00392156862745098
+  },
+  "image_seq_length": 280,
+  "processor_class": "Gemma4Processor",
+  "video_processor": {
+    "do_convert_rgb": true,
+    "do_normalize": true,
+    "do_rescale": true,
+    "do_resize": true,
+    "do_sample_frames": true,
+    "image_mean": [
+      0.0,
+      0.0,
+      0.0
+    ],
+    "image_std": [
+      1.0,
+      1.0,
+      1.0
+    ],
+    "max_soft_tokens": 70,
+    "num_frames": 32,
+    "patch_size": 16,
+    "pooling_kernel_size": 3,
+    "resample": 3,
+    "rescale_factor": 0.00392156862745098,
+    "return_metadata": false,
+    "video_processor_type": "Gemma4VideoProcessor"
+  }
+}

checkpoint-600/rng_state.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3d6d8fafcd1ee268414be5acf0366296af5b03d60871978712eac1979cb42d65
+size 14645

checkpoint-600/scheduler.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d16e41566cd79f9a45bf4dcd02a2eb09de5c752a5b1f5bddaf3d8672c4407ffc
+size 1465

checkpoint-600/tokenizer.json

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:cc8d3a0ce36466ccc1278bf987df5f71db1719b9ca6b4118264f45cb627bfe0f
+size 32169626

checkpoint-600/tokenizer_config.json

@@ -0,0 +1,95 @@
+{
+  "audio_token": "<|audio|>",
+  "backend": "tokenizers",
+  "boa_token": "<|audio>",
+  "boi_token": "<|image>",
+  "bos_token": "<bos>",
+  "eoa_token": "<audio|>",
+  "eoc_token": "<channel|>",
+  "eoi_token": "<image|>",
+  "eos_token": "<turn|>",
+  "eot_token": "<turn|>",
+  "escape_token": "<|\"|>",
+  "etc_token": "<tool_call|>",
+  "etd_token": "<tool|>",
+  "etr_token": "<tool_response|>",
+  "extra_special_tokens": [
+    "<|video|>"
+  ],
+  "image_token": "<|image|>",
+  "is_local": false,
+  "mask_token": "<mask>",
+  "model_max_length": 131072,
+  "model_specific_special_tokens": {
+    "audio_token": "<|audio|>",
+    "boa_token": "<|audio>",
+    "boi_token": "<|image>",
+    "eoa_token": "<audio|>",
+    "eoc_token": "<channel|>",
+    "eoi_token": "<image|>",
+    "eot_token": "<turn|>",
+    "escape_token": "<|\"|>",
+    "etc_token": "<tool_call|>",
+    "etd_token": "<tool|>",
+    "etr_token": "<tool_response|>",
+    "image_token": "<|image|>",
+    "soc_token": "<|channel>",
+    "sot_token": "<|turn>",
+    "stc_token": "<|tool_call>",
+    "std_token": "<|tool>",
+    "str_token": "<|tool_response>",
+    "think_token": "<|think|>"
+  },
+  "pad_token": "<pad>",
+  "padding_side": "right",
+  "processor_class": "Gemma4Processor",
+  "response_schema": {
+    "properties": {
+      "content": {
+        "type": "string"
+      },
+      "role": {
+        "const": "assistant"
+      },
+      "thinking": {
+        "type": "string"
+      },
+      "tool_calls": {
+        "items": {
+          "properties": {
+            "function": {
+              "properties": {
+                "arguments": {
+                  "additionalProperties": {},
+                  "type": "object",
+                  "x-parser": "gemma4-tool-call"
+                },
+                "name": {
+                  "type": "string"
+                }
+              },
+              "type": "object",
+              "x-regex": "call\\:(?P<name>\\w+)(?P<arguments>\\{.*\\})"
+            },
+            "type": {
+              "const": "function"
+            }
+          },
+          "type": "object"
+        },
+        "type": "array",
+        "x-regex-iterator": "<\\|tool_call>(.*?)<tool_call\\|>"
+      }
+    },
+    "type": "object",
+    "x-regex": "(\\<\\|channel\\>thought\\n(?P<thinking>.*?)\\<channel\\|\\>)?(?P<tool_calls>\\<\\|tool_call\\>.*\\<tool_call\\|\\>)?(?P<content>(?:(?!\\<turn\\|\\>)(?!\\<\\|tool_response\\>).)+)?(?:\\<turn\\|\\>|\\<\\|tool_response\\>)?"
+  },
+  "soc_token": "<|channel>",
+  "sot_token": "<|turn>",
+  "stc_token": "<|tool_call>",
+  "std_token": "<|tool>",
+  "str_token": "<|tool_response>",
+  "think_token": "<|think|>",
+  "tokenizer_class": "GemmaTokenizer",
+  "unk_token": "<unk>"
+}

checkpoint-600/training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ae7a694dd4875ee97139bdd8ebc7349b14ce4d529c119642d14f7f7f5ebd9df4
+size 5777

processor_config.json

@@ -0,0 +1,75 @@
+{
+  "audio_ms_per_token": 40,
+  "audio_seq_length": 750,
+  "feature_extractor": {
+    "dither": 0.0,
+    "feature_extractor_type": "Gemma4AudioFeatureExtractor",
+    "feature_size": 128,
+    "fft_length": 512,
+    "fft_overdrive": false,
+    "frame_length": 320,
+    "hop_length": 160,
+    "input_scale_factor": 1.0,
+    "max_frequency": 8000.0,
+    "mel_floor": 0.001,
+    "min_frequency": 0.0,
+    "padding_side": "left",
+    "padding_value": 0.0,
+    "per_bin_mean": null,
+    "per_bin_stddev": null,
+    "preemphasis": 0.0,
+    "preemphasis_htk_flavor": true,
+    "return_attention_mask": true,
+    "sampling_rate": 16000
+  },
+  "image_processor": {
+    "do_convert_rgb": true,
+    "do_normalize": false,
+    "do_rescale": true,
+    "do_resize": true,
+    "image_mean": [
+      0.0,
+      0.0,
+      0.0
+    ],
+    "image_processor_type": "Gemma4ImageProcessor",
+    "image_seq_length": 280,
+    "image_std": [
+      1.0,
+      1.0,
+      1.0
+    ],
+    "max_soft_tokens": 280,
+    "patch_size": 16,
+    "pooling_kernel_size": 3,
+    "resample": 3,
+    "rescale_factor": 0.00392156862745098
+  },
+  "image_seq_length": 280,
+  "processor_class": "Gemma4Processor",
+  "video_processor": {
+    "do_convert_rgb": true,
+    "do_normalize": true,
+    "do_rescale": true,
+    "do_resize": true,
+    "do_sample_frames": true,
+    "image_mean": [
+      0.0,
+      0.0,
+      0.0
+    ],
+    "image_std": [
+      1.0,
+      1.0,
+      1.0
+    ],
+    "max_soft_tokens": 70,
+    "num_frames": 32,
+    "patch_size": 16,
+    "pooling_kernel_size": 3,
+    "resample": 3,
+    "rescale_factor": 0.00392156862745098,
+    "return_metadata": false,
+    "video_processor_type": "Gemma4VideoProcessor"
+  }
+}

tokenizer.json

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:cc8d3a0ce36466ccc1278bf987df5f71db1719b9ca6b4118264f45cb627bfe0f
+size 32169626

tokenizer_config.json

@@ -0,0 +1,95 @@
+{
+  "audio_token": "<|audio|>",
+  "backend": "tokenizers",
+  "boa_token": "<|audio>",
+  "boi_token": "<|image>",
+  "bos_token": "<bos>",
+  "eoa_token": "<audio|>",
+  "eoc_token": "<channel|>",
+  "eoi_token": "<image|>",
+  "eos_token": "<turn|>",
+  "eot_token": "<turn|>",
+  "escape_token": "<|\"|>",
+  "etc_token": "<tool_call|>",
+  "etd_token": "<tool|>",
+  "etr_token": "<tool_response|>",
+  "extra_special_tokens": [
+    "<|video|>"
+  ],
+  "image_token": "<|image|>",
+  "is_local": false,
+  "mask_token": "<mask>",
+  "model_max_length": 131072,
+  "model_specific_special_tokens": {
+    "audio_token": "<|audio|>",
+    "boa_token": "<|audio>",
+    "boi_token": "<|image>",
+    "eoa_token": "<audio|>",
+    "eoc_token": "<channel|>",
+    "eoi_token": "<image|>",
+    "eot_token": "<turn|>",
+    "escape_token": "<|\"|>",
+    "etc_token": "<tool_call|>",
+    "etd_token": "<tool|>",
+    "etr_token": "<tool_response|>",
+    "image_token": "<|image|>",
+    "soc_token": "<|channel>",
+    "sot_token": "<|turn>",
+    "stc_token": "<|tool_call>",
+    "std_token": "<|tool>",
+    "str_token": "<|tool_response>",
+    "think_token": "<|think|>"
+  },
+  "pad_token": "<pad>",
+  "padding_side": "right",
+  "processor_class": "Gemma4Processor",
+  "response_schema": {
+    "properties": {
+      "content": {
+        "type": "string"
+      },
+      "role": {
+        "const": "assistant"
+      },
+      "thinking": {
+        "type": "string"
+      },
+      "tool_calls": {
+        "items": {
+          "properties": {
+            "function": {
+              "properties": {
+                "arguments": {
+                  "additionalProperties": {},
+                  "type": "object",
+                  "x-parser": "gemma4-tool-call"
+                },
+                "name": {
+                  "type": "string"
+                }
+              },
+              "type": "object",
+              "x-regex": "call\\:(?P<name>\\w+)(?P<arguments>\\{.*\\})"
+            },
+            "type": {
+              "const": "function"
+            }
+          },
+          "type": "object"
+        },
+        "type": "array",
+        "x-regex-iterator": "<\\|tool_call>(.*?)<tool_call\\|>"
+      }
+    },
+    "type": "object",
+    "x-regex": "(\\<\\|channel\\>thought\\n(?P<thinking>.*?)\\<channel\\|\\>)?(?P<tool_calls>\\<\\|tool_call\\>.*\\<tool_call\\|\\>)?(?P<content>(?:(?!\\<turn\\|\\>)(?!\\<\\|tool_response\\>).)+)?(?:\\<turn\\|\\>|\\<\\|tool_response\\>)?"
+  },
+  "soc_token": "<|channel>",
+  "sot_token": "<|turn>",
+  "stc_token": "<|tool_call>",
+  "std_token": "<|tool>",
+  "str_token": "<|tool_response>",
+  "think_token": "<|think|>",
+  "tokenizer_class": "GemmaTokenizer",
+  "unk_token": "<unk>"
+}

training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ae7a694dd4875ee97139bdd8ebc7349b14ce4d529c119642d14f7f7f5ebd9df4
+size 5777