diff --git a/README.md b/README.md
index fd5283695762652b485d13f396ca2f4e3cc01bcd..588cf705f719baf7cb6bd2749f5529e9cebec8a2 100644
--- a/README.md
+++ b/README.md
@@ -1,21 +1,63 @@
---
base_model: unsloth/gemma-4-e2b-it-unsloth-bnb-4bit
+library_name: peft
+model_name: unsloth_gemma-4-E2B-it_1780218767
tags:
-- text-generation-inference
+- base_model:adapter:unsloth/gemma-4-e2b-it-unsloth-bnb-4bit
+- lora
+- sft
- transformers
+- trl
- unsloth
-- gemma4
-license: apache-2.0
-language:
-- en
+licence: license
+pipeline_tag: text-generation
---
-# Uploaded finetuned model
+# Model Card for unsloth_gemma-4-E2B-it_1780218767
-- **Developed by:** prog-love
-- **License:** apache-2.0
-- **Finetuned from model :** unsloth/gemma-4-e2b-it-unsloth-bnb-4bit
+This model is a fine-tuned version of [unsloth/gemma-4-e2b-it-unsloth-bnb-4bit](https://huggingface.co/unsloth/gemma-4-e2b-it-unsloth-bnb-4bit).
+It has been trained using [TRL](https://github.com/huggingface/trl).
-This gemma4 model was trained 2x faster with [Unsloth](https://github.com/unslothai/unsloth) and Huggingface's TRL library.
+## Quick start
-[
](https://github.com/unslothai/unsloth)
+```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}}
+}
+```
\ No newline at end of file
diff --git a/adapter_config.json b/adapter_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..78d5b0d5d0dac5f323442e40d7ce2edccb4e3151
--- /dev/null
+++ b/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": "unsloth/gemma-4-e2b-it-unsloth-bnb-4bit",
+ "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": [
+ "q_proj",
+ "gate_proj",
+ "v_proj",
+ "k_proj",
+ "down_proj",
+ "o_proj",
+ "up_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"
+}
\ No newline at end of file
diff --git a/adapter_model.safetensors b/adapter_model.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..2c87a0ef165f037ee646d9506ba63a7bcc4aa079
--- /dev/null
+++ b/adapter_model.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:dae7c101be06c818341a1b3354a5715c93ad40fc4c4700a81f0c71c0e699b82b
+size 124277728
diff --git a/checkpoint-120/README.md b/checkpoint-120/README.md
new file mode 100644
index 0000000000000000000000000000000000000000..41470ec6557bbc37155893f70bd59986401b9148
--- /dev/null
+++ b/checkpoint-120/README.md
@@ -0,0 +1,210 @@
+---
+base_model: unsloth/gemma-4-e2b-it-unsloth-bnb-4bit
+library_name: peft
+pipeline_tag: text-generation
+tags:
+- base_model:adapter:unsloth/gemma-4-e2b-it-unsloth-bnb-4bit
+- lora
+- sft
+- transformers
+- trl
+- unsloth
+---
+
+# Model Card for Model ID
+
+
+
+
+
+## Model Details
+
+### Model Description
+
+
+
+
+
+- **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]
+
+
+
+- **Repository:** [More Information Needed]
+- **Paper [optional]:** [More Information Needed]
+- **Demo [optional]:** [More Information Needed]
+
+## Uses
+
+
+
+### Direct Use
+
+
+
+[More Information Needed]
+
+### Downstream Use [optional]
+
+
+
+[More Information Needed]
+
+### Out-of-Scope Use
+
+
+
+[More Information Needed]
+
+## Bias, Risks, and Limitations
+
+
+
+[More Information Needed]
+
+### Recommendations
+
+
+
+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
+
+
+
+[More Information Needed]
+
+### Training Procedure
+
+
+
+#### Preprocessing [optional]
+
+[More Information Needed]
+
+
+#### Training Hyperparameters
+
+- **Training regime:** [More Information Needed]
+
+#### Speeds, Sizes, Times [optional]
+
+
+
+[More Information Needed]
+
+## Evaluation
+
+
+
+### Testing Data, Factors & Metrics
+
+#### Testing Data
+
+
+
+[More Information Needed]
+
+#### Factors
+
+
+
+[More Information Needed]
+
+#### Metrics
+
+
+
+[More Information Needed]
+
+### Results
+
+[More Information Needed]
+
+#### Summary
+
+
+
+## Model Examination [optional]
+
+
+
+[More Information Needed]
+
+## Environmental Impact
+
+
+
+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]
+
+
+
+**BibTeX:**
+
+[More Information Needed]
+
+**APA:**
+
+[More Information Needed]
+
+## Glossary [optional]
+
+
+
+[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
\ No newline at end of file
diff --git a/checkpoint-120/adapter_config.json b/checkpoint-120/adapter_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..915726be1cd67eb165724a8390b51a57fe0dd6df
--- /dev/null
+++ b/checkpoint-120/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": "unsloth/gemma-4-e2b-it-unsloth-bnb-4bit",
+ "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": [
+ "q_proj",
+ "gate_proj",
+ "v_proj",
+ "k_proj",
+ "down_proj",
+ "o_proj",
+ "up_proj"
+ ],
+ "target_parameters": null,
+ "task_type": "CAUSAL_LM",
+ "trainable_token_indices": null,
+ "use_dora": false,
+ "use_qalora": false,
+ "use_rslora": false
+}
\ No newline at end of file
diff --git a/checkpoint-120/adapter_model.safetensors b/checkpoint-120/adapter_model.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..8f9f4a3dfabb88103f6bd5c376c21e31b851bcfd
--- /dev/null
+++ b/checkpoint-120/adapter_model.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a2fa7f51110b2c95c35000062cb6a564c25910d5c3d47a3ab9455def2a9d031c
+size 124277728
diff --git a/checkpoint-120/chat_template.jinja b/checkpoint-120/chat_template.jinja
new file mode 100644
index 0000000000000000000000000000000000000000..dc032e46e28e0d325b18d26da1279cdda3afa8d7
--- /dev/null
+++ b/checkpoint-120/chat_template.jinja
@@ -0,0 +1,351 @@
+{%- macro format_parameters(properties, required, filter_keys=false) -%}
+ {%- 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 not filter_keys or 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([]), filter_keys=true) -}}
+ }
+ {%- 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('') -%}
+ {%- 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 -%}
+ {{- '' -}}
+{%- 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'] -%}
+ {%- if messages[0]['content'] is string -%}
+ {{- messages[0]['content'] | trim -}}
+ {%- elif messages[0]['content'] is sequence -%}
+ {%- for item in messages[0]['content'] -%}
+ {{- item['text'] | trim + ' '-}}
+ {%- endfor -%}
+ {%- endif -%}
+ {%- set loop_messages = messages[1:] -%}
+ {%- endif -%}
+ {%- if tools -%}
+ {%- for tool in tools %}
+ {{- '<|tool>' -}}
+ {{- format_function_declaration(tool) | trim -}}
+ {{- '' -}}
+ {%- endfor %}
+ {%- set ns.prev_message_type = 'tool' -%}
+ {%- endif -%}
+ {{- '\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' -}}
+ {%- 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 -%}
+ {{- '}' -}}
+ {%- 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 -%}
+
+ {%- set captured_content -%}
+ {%- 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 -%}
+ {%- endset -%}
+
+ {{- captured_content -}}
+ {%- set has_content = captured_content | trim | length > 0 -%}
+
+ {%- if ns.prev_message_type == 'tool_call' and not ns_tr_out.flag -%}
+ {{- '<|tool_response>' -}}
+ {%- elif not (ns_tr_out.flag and not has_content) -%}
+ {{- '\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 -%}
\ No newline at end of file
diff --git a/checkpoint-120/optimizer.pt b/checkpoint-120/optimizer.pt
new file mode 100644
index 0000000000000000000000000000000000000000..0e06b19d377b8fadae90e2505364557f82b3f19b
--- /dev/null
+++ b/checkpoint-120/optimizer.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:095e1fcdf14af0658b82cfeea97af6ef9916881ed1b2df6af4546d0f79623b85
+size 52047117
diff --git a/checkpoint-120/processor_config.json b/checkpoint-120/processor_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..bfaa5e360ad17c2761b7ee3b05abf03e381e8aa1
--- /dev/null
+++ b/checkpoint-120/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"
+ }
+}
diff --git a/checkpoint-120/rng_state.pth b/checkpoint-120/rng_state.pth
new file mode 100644
index 0000000000000000000000000000000000000000..3ef66339b9befa098183fd5d69faed6838e526b0
--- /dev/null
+++ b/checkpoint-120/rng_state.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f1d565802a8e26c4e8a31328752b7a7fdc186d9401aa008e65697d0ad8c22e33
+size 14645
diff --git a/checkpoint-120/scaler.pt b/checkpoint-120/scaler.pt
new file mode 100644
index 0000000000000000000000000000000000000000..3b3554b6c901fe0b965040dde5865974f5d9a42c
--- /dev/null
+++ b/checkpoint-120/scaler.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:aa76f9fdc8f6bd88ee74f70ad294ae36bf2fe178b9965d6d01f072e8fa9198a8
+size 1383
diff --git a/checkpoint-120/scheduler.pt b/checkpoint-120/scheduler.pt
new file mode 100644
index 0000000000000000000000000000000000000000..f95fbe157a2c9324ca9f4be1525b136b26dc5307
--- /dev/null
+++ b/checkpoint-120/scheduler.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2d142e1b585c4fbc6a90a9d8170d8738b8abf95d0d40cf8046076286563427f1
+size 1465
diff --git a/checkpoint-120/tokenizer.json b/checkpoint-120/tokenizer.json
new file mode 100644
index 0000000000000000000000000000000000000000..1ff9f3e3439a939b971f9919e821bf87e835a503
--- /dev/null
+++ b/checkpoint-120/tokenizer.json
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:cc8d3a0ce36466ccc1278bf987df5f71db1719b9ca6b4118264f45cb627bfe0f
+size 32169626
diff --git a/checkpoint-120/tokenizer_config.json b/checkpoint-120/tokenizer_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..24b1522bb32b0eb05b631b1a9f112751b8cb0fab
--- /dev/null
+++ b/checkpoint-120/tokenizer_config.json
@@ -0,0 +1,289 @@
+{
+ "audio_token": "<|audio|>",
+ "backend": "tokenizers",
+ "boa_token": "<|audio>",
+ "boi_token": "<|image>",
+ "bos_token": "",
+ "eoa_token": "",
+ "eoc_token": "",
+ "eoi_token": "",
+ "eos_token": "",
+ "eot_token": "",
+ "escape_token": "<|\"|>",
+ "etc_token": "",
+ "etd_token": "",
+ "etr_token": "",
+ "extra_special_tokens": [
+ "<|video|>"
+ ],
+ "image_token": "<|image|>",
+ "is_local": false,
+ "mask_token": "",
+ "model_max_length": 131072,
+ "model_specific_special_tokens": {
+ "audio_token": "<|audio|>",
+ "boa_token": "<|audio>",
+ "boi_token": "<|image>",
+ "eoa_token": "",
+ "eoc_token": "",
+ "eoi_token": "",
+ "eot_token": "",
+ "escape_token": "<|\"|>",
+ "etc_token": "",
+ "etd_token": "",
+ "etr_token": "",
+ "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": "",
+ "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\\w+)(?P\\{.*\\})"
+ },
+ "type": {
+ "const": "function"
+ }
+ },
+ "type": "object"
+ },
+ "type": "array",
+ "x-regex-iterator": "<\\|tool_call>(.*?)"
+ }
+ },
+ "type": "object",
+ "x-regex": "(\\<\\|channel\\>thought\\n(?P.*?)\\)?(?P\\<\\|tool_call\\>.*\\)?(?P(?:(?!\\)(?!\\<\\|tool_response\\>).)+)?(?:\\|\\<\\|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": "",
+ "added_tokens_decoder": {
+ "0": {
+ "content": "",
+ "single_word": false,
+ "lstrip": false,
+ "rstrip": false,
+ "normalized": false,
+ "special": true
+ },
+ "1": {
+ "content": "",
+ "single_word": false,
+ "lstrip": false,
+ "rstrip": false,
+ "normalized": false,
+ "special": true
+ },
+ "2": {
+ "content": "",
+ "single_word": false,
+ "lstrip": false,
+ "rstrip": false,
+ "normalized": false,
+ "special": true
+ },
+ "3": {
+ "content": "",
+ "single_word": false,
+ "lstrip": false,
+ "rstrip": false,
+ "normalized": false,
+ "special": true
+ },
+ "4": {
+ "content": "",
+ "single_word": false,
+ "lstrip": false,
+ "rstrip": false,
+ "normalized": false,
+ "special": true
+ },
+ "46": {
+ "content": "<|tool>",
+ "single_word": false,
+ "lstrip": false,
+ "rstrip": false,
+ "normalized": false,
+ "special": true
+ },
+ "47": {
+ "content": "",
+ "single_word": false,
+ "lstrip": false,
+ "rstrip": false,
+ "normalized": false,
+ "special": true
+ },
+ "48": {
+ "content": "<|tool_call>",
+ "single_word": false,
+ "lstrip": false,
+ "rstrip": false,
+ "normalized": false,
+ "special": true
+ },
+ "49": {
+ "content": "",
+ "single_word": false,
+ "lstrip": false,
+ "rstrip": false,
+ "normalized": false,
+ "special": true
+ },
+ "50": {
+ "content": "<|tool_response>",
+ "single_word": false,
+ "lstrip": false,
+ "rstrip": false,
+ "normalized": false,
+ "special": true
+ },
+ "51": {
+ "content": "",
+ "single_word": false,
+ "lstrip": false,
+ "rstrip": false,
+ "normalized": false,
+ "special": true
+ },
+ "52": {
+ "content": "<|\"|>",
+ "single_word": false,
+ "lstrip": false,
+ "rstrip": false,
+ "normalized": false,
+ "special": true
+ },
+ "98": {
+ "content": "<|think|>",
+ "single_word": false,
+ "lstrip": false,
+ "rstrip": false,
+ "normalized": false,
+ "special": true
+ },
+ "100": {
+ "content": "<|channel>",
+ "single_word": false,
+ "lstrip": false,
+ "rstrip": false,
+ "normalized": false,
+ "special": true
+ },
+ "101": {
+ "content": "",
+ "single_word": false,
+ "lstrip": false,
+ "rstrip": false,
+ "normalized": false,
+ "special": true
+ },
+ "105": {
+ "content": "<|turn>",
+ "single_word": false,
+ "lstrip": false,
+ "rstrip": false,
+ "normalized": false,
+ "special": true
+ },
+ "106": {
+ "content": "",
+ "single_word": false,
+ "lstrip": false,
+ "rstrip": false,
+ "normalized": false,
+ "special": true
+ },
+ "255999": {
+ "content": "<|image>",
+ "single_word": false,
+ "lstrip": false,
+ "rstrip": false,
+ "normalized": false,
+ "special": true
+ },
+ "256000": {
+ "content": "<|audio>",
+ "single_word": false,
+ "lstrip": false,
+ "rstrip": false,
+ "normalized": false,
+ "special": true
+ },
+ "258880": {
+ "content": "<|image|>",
+ "single_word": false,
+ "lstrip": false,
+ "rstrip": false,
+ "normalized": false,
+ "special": true
+ },
+ "258881": {
+ "content": "<|audio|>",
+ "single_word": false,
+ "lstrip": false,
+ "rstrip": false,
+ "normalized": false,
+ "special": true
+ },
+ "258882": {
+ "content": "",
+ "single_word": false,
+ "lstrip": false,
+ "rstrip": false,
+ "normalized": false,
+ "special": true
+ },
+ "258883": {
+ "content": "",
+ "single_word": false,
+ "lstrip": false,
+ "rstrip": false,
+ "normalized": false,
+ "special": true
+ },
+ "258884": {
+ "content": "<|video|>",
+ "single_word": false,
+ "lstrip": false,
+ "rstrip": false,
+ "normalized": false,
+ "special": true
+ }
+ }
+}
diff --git a/checkpoint-120/trainer_state.json b/checkpoint-120/trainer_state.json
new file mode 100644
index 0000000000000000000000000000000000000000..f00bc6275b952c8f9bea0301731173e0478935d7
--- /dev/null
+++ b/checkpoint-120/trainer_state.json
@@ -0,0 +1,1234 @@
+{
+ "best_global_step": null,
+ "best_metric": null,
+ "best_model_checkpoint": null,
+ "epoch": 0.028402366863905324,
+ "eval_steps": 500,
+ "global_step": 120,
+ "is_hyper_param_search": false,
+ "is_local_process_zero": true,
+ "is_world_process_zero": true,
+ "log_history": [
+ {
+ "epoch": 0.00023668639053254438,
+ "grad_norm": NaN,
+ "learning_rate": 0.0,
+ "loss": 4.347962379455566,
+ "num_input_tokens_seen": 842,
+ "step": 1,
+ "train_runtime": 63.0289,
+ "train_tokens_per_second": 13.359
+ },
+ {
+ "epoch": 0.00047337278106508875,
+ "grad_norm": NaN,
+ "learning_rate": 0.0,
+ "loss": 4.634688854217529,
+ "num_input_tokens_seen": 1820,
+ "step": 2,
+ "train_runtime": 67.7033,
+ "train_tokens_per_second": 26.882
+ },
+ {
+ "epoch": 0.0007100591715976331,
+ "grad_norm": 13.944212913513184,
+ "learning_rate": 0.0,
+ "loss": 4.42241096496582,
+ "num_input_tokens_seen": 2754,
+ "step": 3,
+ "train_runtime": 72.324,
+ "train_tokens_per_second": 38.079
+ },
+ {
+ "epoch": 0.0009467455621301775,
+ "grad_norm": 13.445413589477539,
+ "learning_rate": 4e-05,
+ "loss": 4.682804107666016,
+ "num_input_tokens_seen": 3700,
+ "step": 4,
+ "train_runtime": 75.3925,
+ "train_tokens_per_second": 49.077
+ },
+ {
+ "epoch": 0.001183431952662722,
+ "grad_norm": 10.576727867126465,
+ "learning_rate": 8e-05,
+ "loss": 4.166159629821777,
+ "num_input_tokens_seen": 4714,
+ "step": 5,
+ "train_runtime": 78.474,
+ "train_tokens_per_second": 60.071
+ },
+ {
+ "epoch": 0.0014201183431952662,
+ "grad_norm": 8.077850341796875,
+ "learning_rate": 0.00012,
+ "loss": 3.2006261348724365,
+ "num_input_tokens_seen": 5674,
+ "step": 6,
+ "train_runtime": 81.6139,
+ "train_tokens_per_second": 69.522
+ },
+ {
+ "epoch": 0.0016568047337278107,
+ "grad_norm": 6.60789680480957,
+ "learning_rate": 0.00016,
+ "loss": 2.7067415714263916,
+ "num_input_tokens_seen": 6646,
+ "step": 7,
+ "train_runtime": 84.786,
+ "train_tokens_per_second": 78.386
+ },
+ {
+ "epoch": 0.001893491124260355,
+ "grad_norm": 9.175209045410156,
+ "learning_rate": 0.0002,
+ "loss": 2.367570638656616,
+ "num_input_tokens_seen": 7518,
+ "step": 8,
+ "train_runtime": 87.8178,
+ "train_tokens_per_second": 85.609
+ },
+ {
+ "epoch": 0.0021301775147928993,
+ "grad_norm": 3.438948631286621,
+ "learning_rate": 0.00019891891891891895,
+ "loss": 2.1013119220733643,
+ "num_input_tokens_seen": 8470,
+ "step": 9,
+ "train_runtime": 90.9458,
+ "train_tokens_per_second": 93.132
+ },
+ {
+ "epoch": 0.002366863905325444,
+ "grad_norm": 3.459472894668579,
+ "learning_rate": 0.00019783783783783784,
+ "loss": 1.8991940021514893,
+ "num_input_tokens_seen": 9458,
+ "step": 10,
+ "train_runtime": 94.0577,
+ "train_tokens_per_second": 100.555
+ },
+ {
+ "epoch": 0.0026035502958579883,
+ "grad_norm": 2.169750928878784,
+ "learning_rate": 0.00019675675675675678,
+ "loss": 1.7395055294036865,
+ "num_input_tokens_seen": 10456,
+ "step": 11,
+ "train_runtime": 97.2068,
+ "train_tokens_per_second": 107.565
+ },
+ {
+ "epoch": 0.0028402366863905324,
+ "grad_norm": 1.8638927936553955,
+ "learning_rate": 0.00019567567567567567,
+ "loss": 1.5622950792312622,
+ "num_input_tokens_seen": 11410,
+ "step": 12,
+ "train_runtime": 100.2911,
+ "train_tokens_per_second": 113.769
+ },
+ {
+ "epoch": 0.003076923076923077,
+ "grad_norm": 1.4741116762161255,
+ "learning_rate": 0.00019459459459459462,
+ "loss": 1.392699956893921,
+ "num_input_tokens_seen": 12316,
+ "step": 13,
+ "train_runtime": 103.3664,
+ "train_tokens_per_second": 119.149
+ },
+ {
+ "epoch": 0.0033136094674556214,
+ "grad_norm": 1.3238699436187744,
+ "learning_rate": 0.00019351351351351353,
+ "loss": 1.2535074949264526,
+ "num_input_tokens_seen": 13256,
+ "step": 14,
+ "train_runtime": 106.4031,
+ "train_tokens_per_second": 124.583
+ },
+ {
+ "epoch": 0.0035502958579881655,
+ "grad_norm": 1.276484489440918,
+ "learning_rate": 0.00019243243243243245,
+ "loss": 1.135048508644104,
+ "num_input_tokens_seen": 14228,
+ "step": 15,
+ "train_runtime": 109.4909,
+ "train_tokens_per_second": 129.947
+ },
+ {
+ "epoch": 0.00378698224852071,
+ "grad_norm": 1.1560865640640259,
+ "learning_rate": 0.00019135135135135137,
+ "loss": 1.0317097902297974,
+ "num_input_tokens_seen": 15192,
+ "step": 16,
+ "train_runtime": 112.5741,
+ "train_tokens_per_second": 134.951
+ },
+ {
+ "epoch": 0.004023668639053254,
+ "grad_norm": 1.1331353187561035,
+ "learning_rate": 0.00019027027027027028,
+ "loss": 0.9967219233512878,
+ "num_input_tokens_seen": 16130,
+ "step": 17,
+ "train_runtime": 115.6495,
+ "train_tokens_per_second": 139.473
+ },
+ {
+ "epoch": 0.004260355029585799,
+ "grad_norm": 1.1567702293395996,
+ "learning_rate": 0.0001891891891891892,
+ "loss": 0.905368447303772,
+ "num_input_tokens_seen": 17078,
+ "step": 18,
+ "train_runtime": 118.6796,
+ "train_tokens_per_second": 143.9
+ },
+ {
+ "epoch": 0.004497041420118343,
+ "grad_norm": 0.9157600998878479,
+ "learning_rate": 0.00018810810810810814,
+ "loss": 0.8075690865516663,
+ "num_input_tokens_seen": 18062,
+ "step": 19,
+ "train_runtime": 121.8602,
+ "train_tokens_per_second": 148.219
+ },
+ {
+ "epoch": 0.004733727810650888,
+ "grad_norm": 0.8306605815887451,
+ "learning_rate": 0.00018702702702702703,
+ "loss": 0.8007175922393799,
+ "num_input_tokens_seen": 19078,
+ "step": 20,
+ "train_runtime": 125.0019,
+ "train_tokens_per_second": 152.622
+ },
+ {
+ "epoch": 0.004970414201183432,
+ "grad_norm": 0.690018355846405,
+ "learning_rate": 0.00018594594594594597,
+ "loss": 0.7810791730880737,
+ "num_input_tokens_seen": 20068,
+ "step": 21,
+ "train_runtime": 128.0932,
+ "train_tokens_per_second": 156.667
+ },
+ {
+ "epoch": 0.005207100591715977,
+ "grad_norm": 1.4768867492675781,
+ "learning_rate": 0.00018486486486486486,
+ "loss": 0.7172120809555054,
+ "num_input_tokens_seen": 20972,
+ "step": 22,
+ "train_runtime": 131.0901,
+ "train_tokens_per_second": 159.982
+ },
+ {
+ "epoch": 0.00544378698224852,
+ "grad_norm": 0.7999809980392456,
+ "learning_rate": 0.0001837837837837838,
+ "loss": 0.6333726048469543,
+ "num_input_tokens_seen": 21900,
+ "step": 23,
+ "train_runtime": 134.212,
+ "train_tokens_per_second": 163.175
+ },
+ {
+ "epoch": 0.005680473372781065,
+ "grad_norm": 0.9285996556282043,
+ "learning_rate": 0.00018270270270270272,
+ "loss": 0.6750729084014893,
+ "num_input_tokens_seen": 22770,
+ "step": 24,
+ "train_runtime": 137.3193,
+ "train_tokens_per_second": 165.818
+ },
+ {
+ "epoch": 0.005917159763313609,
+ "grad_norm": 0.7733820080757141,
+ "learning_rate": 0.00018162162162162164,
+ "loss": 0.672032356262207,
+ "num_input_tokens_seen": 23746,
+ "step": 25,
+ "train_runtime": 140.3175,
+ "train_tokens_per_second": 169.23
+ },
+ {
+ "epoch": 0.006153846153846154,
+ "grad_norm": 0.73944091796875,
+ "learning_rate": 0.00018054054054054055,
+ "loss": 0.6716101169586182,
+ "num_input_tokens_seen": 24716,
+ "step": 26,
+ "train_runtime": 143.3683,
+ "train_tokens_per_second": 172.395
+ },
+ {
+ "epoch": 0.006390532544378698,
+ "grad_norm": 0.7659372091293335,
+ "learning_rate": 0.00017945945945945947,
+ "loss": 0.6067175269126892,
+ "num_input_tokens_seen": 25724,
+ "step": 27,
+ "train_runtime": 146.5377,
+ "train_tokens_per_second": 175.545
+ },
+ {
+ "epoch": 0.006627218934911243,
+ "grad_norm": 0.5428617000579834,
+ "learning_rate": 0.00017837837837837839,
+ "loss": 0.6083395481109619,
+ "num_input_tokens_seen": 26712,
+ "step": 28,
+ "train_runtime": 149.6429,
+ "train_tokens_per_second": 178.505
+ },
+ {
+ "epoch": 0.006863905325443787,
+ "grad_norm": 0.5915114283561707,
+ "learning_rate": 0.00017729729729729733,
+ "loss": 0.6463742256164551,
+ "num_input_tokens_seen": 27704,
+ "step": 29,
+ "train_runtime": 152.7501,
+ "train_tokens_per_second": 181.368
+ },
+ {
+ "epoch": 0.007100591715976331,
+ "grad_norm": 1.3350757360458374,
+ "learning_rate": 0.00017621621621621622,
+ "loss": 0.5705698132514954,
+ "num_input_tokens_seen": 28648,
+ "step": 30,
+ "train_runtime": 155.8918,
+ "train_tokens_per_second": 183.768
+ },
+ {
+ "epoch": 0.0073372781065088755,
+ "grad_norm": 0.68421471118927,
+ "learning_rate": 0.00017513513513513516,
+ "loss": 0.5660746097564697,
+ "num_input_tokens_seen": 29622,
+ "step": 31,
+ "train_runtime": 163.3209,
+ "train_tokens_per_second": 181.373
+ },
+ {
+ "epoch": 0.00757396449704142,
+ "grad_norm": 0.5762604475021362,
+ "learning_rate": 0.00017405405405405405,
+ "loss": 0.5483309030532837,
+ "num_input_tokens_seen": 30534,
+ "step": 32,
+ "train_runtime": 166.5185,
+ "train_tokens_per_second": 183.367
+ },
+ {
+ "epoch": 0.0078106508875739646,
+ "grad_norm": 0.5751784443855286,
+ "learning_rate": 0.000172972972972973,
+ "loss": 0.5207005739212036,
+ "num_input_tokens_seen": 31458,
+ "step": 33,
+ "train_runtime": 169.7055,
+ "train_tokens_per_second": 185.368
+ },
+ {
+ "epoch": 0.008047337278106508,
+ "grad_norm": 0.5752039551734924,
+ "learning_rate": 0.0001718918918918919,
+ "loss": 0.5534584522247314,
+ "num_input_tokens_seen": 32424,
+ "step": 34,
+ "train_runtime": 172.9184,
+ "train_tokens_per_second": 187.51
+ },
+ {
+ "epoch": 0.008284023668639054,
+ "grad_norm": 0.5918363332748413,
+ "learning_rate": 0.00017081081081081083,
+ "loss": 0.482683002948761,
+ "num_input_tokens_seen": 33398,
+ "step": 35,
+ "train_runtime": 176.2158,
+ "train_tokens_per_second": 189.529
+ },
+ {
+ "epoch": 0.008520710059171597,
+ "grad_norm": 0.6921846270561218,
+ "learning_rate": 0.00016972972972972974,
+ "loss": 0.46112507581710815,
+ "num_input_tokens_seen": 34324,
+ "step": 36,
+ "train_runtime": 179.3173,
+ "train_tokens_per_second": 191.415
+ },
+ {
+ "epoch": 0.008757396449704143,
+ "grad_norm": 0.6429721117019653,
+ "learning_rate": 0.00016864864864864866,
+ "loss": 0.45629197359085083,
+ "num_input_tokens_seen": 35270,
+ "step": 37,
+ "train_runtime": 182.3558,
+ "train_tokens_per_second": 193.413
+ },
+ {
+ "epoch": 0.008994082840236686,
+ "grad_norm": 0.5477020740509033,
+ "learning_rate": 0.00016756756756756757,
+ "loss": 0.4547877907752991,
+ "num_input_tokens_seen": 36206,
+ "step": 38,
+ "train_runtime": 185.4483,
+ "train_tokens_per_second": 195.235
+ },
+ {
+ "epoch": 0.009230769230769232,
+ "grad_norm": 0.6158634424209595,
+ "learning_rate": 0.00016648648648648652,
+ "loss": 0.367482990026474,
+ "num_input_tokens_seen": 37080,
+ "step": 39,
+ "train_runtime": 188.5114,
+ "train_tokens_per_second": 196.699
+ },
+ {
+ "epoch": 0.009467455621301775,
+ "grad_norm": 0.5405091643333435,
+ "learning_rate": 0.0001654054054054054,
+ "loss": 0.42049410939216614,
+ "num_input_tokens_seen": 38030,
+ "step": 40,
+ "train_runtime": 191.5986,
+ "train_tokens_per_second": 198.488
+ },
+ {
+ "epoch": 0.009704142011834319,
+ "grad_norm": 0.6389183402061462,
+ "learning_rate": 0.00016432432432432435,
+ "loss": 0.3935014307498932,
+ "num_input_tokens_seen": 38942,
+ "step": 41,
+ "train_runtime": 194.8426,
+ "train_tokens_per_second": 199.864
+ },
+ {
+ "epoch": 0.009940828402366864,
+ "grad_norm": 0.6742956042289734,
+ "learning_rate": 0.00016324324324324324,
+ "loss": 0.4373692274093628,
+ "num_input_tokens_seen": 39902,
+ "step": 42,
+ "train_runtime": 197.8938,
+ "train_tokens_per_second": 201.633
+ },
+ {
+ "epoch": 0.010177514792899408,
+ "grad_norm": 0.6083794236183167,
+ "learning_rate": 0.00016216216216216218,
+ "loss": 0.3776765465736389,
+ "num_input_tokens_seen": 40858,
+ "step": 43,
+ "train_runtime": 200.9708,
+ "train_tokens_per_second": 203.303
+ },
+ {
+ "epoch": 0.010414201183431953,
+ "grad_norm": 0.5469992160797119,
+ "learning_rate": 0.0001610810810810811,
+ "loss": 0.31133151054382324,
+ "num_input_tokens_seen": 41760,
+ "step": 44,
+ "train_runtime": 204.1304,
+ "train_tokens_per_second": 204.575
+ },
+ {
+ "epoch": 0.010650887573964497,
+ "grad_norm": 0.6452498435974121,
+ "learning_rate": 0.00016,
+ "loss": 0.42721521854400635,
+ "num_input_tokens_seen": 42740,
+ "step": 45,
+ "train_runtime": 207.1845,
+ "train_tokens_per_second": 206.29
+ },
+ {
+ "epoch": 0.01088757396449704,
+ "grad_norm": 0.5988945960998535,
+ "learning_rate": 0.00015891891891891893,
+ "loss": 0.41375258564949036,
+ "num_input_tokens_seen": 43750,
+ "step": 46,
+ "train_runtime": 210.2912,
+ "train_tokens_per_second": 208.045
+ },
+ {
+ "epoch": 0.011124260355029586,
+ "grad_norm": 0.6452038288116455,
+ "learning_rate": 0.00015783783783783785,
+ "loss": 0.3485276699066162,
+ "num_input_tokens_seen": 44690,
+ "step": 47,
+ "train_runtime": 213.3684,
+ "train_tokens_per_second": 209.45
+ },
+ {
+ "epoch": 0.01136094674556213,
+ "grad_norm": 0.7041454911231995,
+ "learning_rate": 0.00015675675675675676,
+ "loss": 0.3296637535095215,
+ "num_input_tokens_seen": 45666,
+ "step": 48,
+ "train_runtime": 216.4459,
+ "train_tokens_per_second": 210.981
+ },
+ {
+ "epoch": 0.011597633136094675,
+ "grad_norm": 0.6726324558258057,
+ "learning_rate": 0.0001556756756756757,
+ "loss": 0.3179457187652588,
+ "num_input_tokens_seen": 46560,
+ "step": 49,
+ "train_runtime": 219.4697,
+ "train_tokens_per_second": 212.148
+ },
+ {
+ "epoch": 0.011834319526627219,
+ "grad_norm": 0.7513703107833862,
+ "learning_rate": 0.0001545945945945946,
+ "loss": 0.3722391426563263,
+ "num_input_tokens_seen": 47528,
+ "step": 50,
+ "train_runtime": 222.596,
+ "train_tokens_per_second": 213.517
+ },
+ {
+ "epoch": 0.012071005917159764,
+ "grad_norm": 0.6454160213470459,
+ "learning_rate": 0.00015351351351351354,
+ "loss": 0.283376544713974,
+ "num_input_tokens_seen": 48468,
+ "step": 51,
+ "train_runtime": 225.7262,
+ "train_tokens_per_second": 214.72
+ },
+ {
+ "epoch": 0.012307692307692308,
+ "grad_norm": 0.9116244316101074,
+ "learning_rate": 0.00015243243243243243,
+ "loss": 0.4286752939224243,
+ "num_input_tokens_seen": 49444,
+ "step": 52,
+ "train_runtime": 228.803,
+ "train_tokens_per_second": 216.099
+ },
+ {
+ "epoch": 0.012544378698224851,
+ "grad_norm": 0.6157099604606628,
+ "learning_rate": 0.00015135135135135137,
+ "loss": 0.27355721592903137,
+ "num_input_tokens_seen": 50424,
+ "step": 53,
+ "train_runtime": 231.8413,
+ "train_tokens_per_second": 217.494
+ },
+ {
+ "epoch": 0.012781065088757397,
+ "grad_norm": 0.6335808038711548,
+ "learning_rate": 0.00015027027027027028,
+ "loss": 0.37197017669677734,
+ "num_input_tokens_seen": 51328,
+ "step": 54,
+ "train_runtime": 234.9265,
+ "train_tokens_per_second": 218.485
+ },
+ {
+ "epoch": 0.01301775147928994,
+ "grad_norm": 0.6838424205780029,
+ "learning_rate": 0.0001491891891891892,
+ "loss": 0.3504602909088135,
+ "num_input_tokens_seen": 52274,
+ "step": 55,
+ "train_runtime": 238.0114,
+ "train_tokens_per_second": 219.628
+ },
+ {
+ "epoch": 0.013254437869822486,
+ "grad_norm": 0.6566418409347534,
+ "learning_rate": 0.00014810810810810812,
+ "loss": 0.3189927935600281,
+ "num_input_tokens_seen": 53214,
+ "step": 56,
+ "train_runtime": 241.0892,
+ "train_tokens_per_second": 220.723
+ },
+ {
+ "epoch": 0.01349112426035503,
+ "grad_norm": 0.6112971901893616,
+ "learning_rate": 0.00014702702702702703,
+ "loss": 0.3080369830131531,
+ "num_input_tokens_seen": 54162,
+ "step": 57,
+ "train_runtime": 244.2461,
+ "train_tokens_per_second": 221.752
+ },
+ {
+ "epoch": 0.013727810650887575,
+ "grad_norm": 0.6469757556915283,
+ "learning_rate": 0.00014594594594594595,
+ "loss": 0.24279898405075073,
+ "num_input_tokens_seen": 55158,
+ "step": 58,
+ "train_runtime": 247.4064,
+ "train_tokens_per_second": 222.945
+ },
+ {
+ "epoch": 0.013964497041420118,
+ "grad_norm": 0.7099384665489197,
+ "learning_rate": 0.0001448648648648649,
+ "loss": 0.31188705563545227,
+ "num_input_tokens_seen": 56096,
+ "step": 59,
+ "train_runtime": 250.4708,
+ "train_tokens_per_second": 223.962
+ },
+ {
+ "epoch": 0.014201183431952662,
+ "grad_norm": 0.5927040576934814,
+ "learning_rate": 0.00014378378378378378,
+ "loss": 0.25696688890457153,
+ "num_input_tokens_seen": 57012,
+ "step": 60,
+ "train_runtime": 253.7131,
+ "train_tokens_per_second": 224.711
+ },
+ {
+ "epoch": 0.014437869822485207,
+ "grad_norm": 0.5483338236808777,
+ "learning_rate": 0.00014270270270270272,
+ "loss": 0.2725352644920349,
+ "num_input_tokens_seen": 58036,
+ "step": 61,
+ "train_runtime": 260.2844,
+ "train_tokens_per_second": 222.971
+ },
+ {
+ "epoch": 0.014674556213017751,
+ "grad_norm": 0.6806847453117371,
+ "learning_rate": 0.00014162162162162161,
+ "loss": 0.31959983706474304,
+ "num_input_tokens_seen": 59026,
+ "step": 62,
+ "train_runtime": 263.3802,
+ "train_tokens_per_second": 224.109
+ },
+ {
+ "epoch": 0.014911242603550296,
+ "grad_norm": 0.6916660070419312,
+ "learning_rate": 0.00014054054054054056,
+ "loss": 0.312305748462677,
+ "num_input_tokens_seen": 59982,
+ "step": 63,
+ "train_runtime": 266.3794,
+ "train_tokens_per_second": 225.175
+ },
+ {
+ "epoch": 0.01514792899408284,
+ "grad_norm": 0.5516266226768494,
+ "learning_rate": 0.00013945945945945947,
+ "loss": 0.2653953433036804,
+ "num_input_tokens_seen": 60926,
+ "step": 64,
+ "train_runtime": 269.3893,
+ "train_tokens_per_second": 226.163
+ },
+ {
+ "epoch": 0.015384615384615385,
+ "grad_norm": 0.5579270720481873,
+ "learning_rate": 0.0001383783783783784,
+ "loss": 0.20462818443775177,
+ "num_input_tokens_seen": 61844,
+ "step": 65,
+ "train_runtime": 272.3982,
+ "train_tokens_per_second": 227.035
+ },
+ {
+ "epoch": 0.015621301775147929,
+ "grad_norm": 0.7645660638809204,
+ "learning_rate": 0.0001372972972972973,
+ "loss": 0.2650893032550812,
+ "num_input_tokens_seen": 62766,
+ "step": 66,
+ "train_runtime": 275.404,
+ "train_tokens_per_second": 227.905
+ },
+ {
+ "epoch": 0.015857988165680473,
+ "grad_norm": 0.5974738001823425,
+ "learning_rate": 0.00013621621621621622,
+ "loss": 0.21247778832912445,
+ "num_input_tokens_seen": 63690,
+ "step": 67,
+ "train_runtime": 278.4021,
+ "train_tokens_per_second": 228.77
+ },
+ {
+ "epoch": 0.016094674556213016,
+ "grad_norm": 0.5831193327903748,
+ "learning_rate": 0.00013513513513513514,
+ "loss": 0.2024172842502594,
+ "num_input_tokens_seen": 64496,
+ "step": 68,
+ "train_runtime": 281.4313,
+ "train_tokens_per_second": 229.171
+ },
+ {
+ "epoch": 0.016331360946745564,
+ "grad_norm": 0.8678795695304871,
+ "learning_rate": 0.00013405405405405408,
+ "loss": 0.31490638852119446,
+ "num_input_tokens_seen": 65460,
+ "step": 69,
+ "train_runtime": 284.6137,
+ "train_tokens_per_second": 229.996
+ },
+ {
+ "epoch": 0.016568047337278107,
+ "grad_norm": 0.7611344456672668,
+ "learning_rate": 0.00013297297297297297,
+ "loss": 0.24661925435066223,
+ "num_input_tokens_seen": 66374,
+ "step": 70,
+ "train_runtime": 287.7742,
+ "train_tokens_per_second": 230.646
+ },
+ {
+ "epoch": 0.01680473372781065,
+ "grad_norm": 0.6685308814048767,
+ "learning_rate": 0.0001318918918918919,
+ "loss": 0.1843535602092743,
+ "num_input_tokens_seen": 67112,
+ "step": 71,
+ "train_runtime": 290.9321,
+ "train_tokens_per_second": 230.679
+ },
+ {
+ "epoch": 0.017041420118343194,
+ "grad_norm": 0.6150883436203003,
+ "learning_rate": 0.0001308108108108108,
+ "loss": 0.23372295498847961,
+ "num_input_tokens_seen": 68154,
+ "step": 72,
+ "train_runtime": 294.0809,
+ "train_tokens_per_second": 231.753
+ },
+ {
+ "epoch": 0.017278106508875738,
+ "grad_norm": 0.6938812732696533,
+ "learning_rate": 0.00012972972972972974,
+ "loss": 0.22619186341762543,
+ "num_input_tokens_seen": 69102,
+ "step": 73,
+ "train_runtime": 297.1843,
+ "train_tokens_per_second": 232.522
+ },
+ {
+ "epoch": 0.017514792899408285,
+ "grad_norm": 0.617787778377533,
+ "learning_rate": 0.00012864864864864866,
+ "loss": 0.2535328269004822,
+ "num_input_tokens_seen": 70010,
+ "step": 74,
+ "train_runtime": 300.2851,
+ "train_tokens_per_second": 233.145
+ },
+ {
+ "epoch": 0.01775147928994083,
+ "grad_norm": 0.6884900331497192,
+ "learning_rate": 0.00012756756756756758,
+ "loss": 0.21632739901542664,
+ "num_input_tokens_seen": 70940,
+ "step": 75,
+ "train_runtime": 303.3533,
+ "train_tokens_per_second": 233.853
+ },
+ {
+ "epoch": 0.017988165680473372,
+ "grad_norm": 0.5782189965248108,
+ "learning_rate": 0.0001264864864864865,
+ "loss": 0.20167675614356995,
+ "num_input_tokens_seen": 71862,
+ "step": 76,
+ "train_runtime": 306.4503,
+ "train_tokens_per_second": 234.498
+ },
+ {
+ "epoch": 0.018224852071005916,
+ "grad_norm": 0.6541483402252197,
+ "learning_rate": 0.0001254054054054054,
+ "loss": 0.23231221735477448,
+ "num_input_tokens_seen": 72812,
+ "step": 77,
+ "train_runtime": 309.5743,
+ "train_tokens_per_second": 235.2
+ },
+ {
+ "epoch": 0.018461538461538463,
+ "grad_norm": 0.7524886727333069,
+ "learning_rate": 0.00012432432432432433,
+ "loss": 0.27632662653923035,
+ "num_input_tokens_seen": 73774,
+ "step": 78,
+ "train_runtime": 312.6963,
+ "train_tokens_per_second": 235.929
+ },
+ {
+ "epoch": 0.018698224852071007,
+ "grad_norm": 0.716282308101654,
+ "learning_rate": 0.00012324324324324327,
+ "loss": 0.2734939754009247,
+ "num_input_tokens_seen": 74716,
+ "step": 79,
+ "train_runtime": 315.7612,
+ "train_tokens_per_second": 236.622
+ },
+ {
+ "epoch": 0.01893491124260355,
+ "grad_norm": 0.7010631561279297,
+ "learning_rate": 0.00012216216216216216,
+ "loss": 0.2774362564086914,
+ "num_input_tokens_seen": 75692,
+ "step": 80,
+ "train_runtime": 318.7983,
+ "train_tokens_per_second": 237.429
+ },
+ {
+ "epoch": 0.019171597633136094,
+ "grad_norm": 0.6561240553855896,
+ "learning_rate": 0.00012108108108108109,
+ "loss": 0.21640686690807343,
+ "num_input_tokens_seen": 76558,
+ "step": 81,
+ "train_runtime": 321.889,
+ "train_tokens_per_second": 237.84
+ },
+ {
+ "epoch": 0.019408284023668638,
+ "grad_norm": 0.5552831888198853,
+ "learning_rate": 0.00012,
+ "loss": 0.21232956647872925,
+ "num_input_tokens_seen": 77408,
+ "step": 82,
+ "train_runtime": 325.0469,
+ "train_tokens_per_second": 238.144
+ },
+ {
+ "epoch": 0.019644970414201185,
+ "grad_norm": 0.5905120372772217,
+ "learning_rate": 0.00011891891891891893,
+ "loss": 0.22443550825119019,
+ "num_input_tokens_seen": 78278,
+ "step": 83,
+ "train_runtime": 328.099,
+ "train_tokens_per_second": 238.58
+ },
+ {
+ "epoch": 0.01988165680473373,
+ "grad_norm": 0.6606366634368896,
+ "learning_rate": 0.00011783783783783784,
+ "loss": 0.2400747388601303,
+ "num_input_tokens_seen": 79262,
+ "step": 84,
+ "train_runtime": 331.1361,
+ "train_tokens_per_second": 239.364
+ },
+ {
+ "epoch": 0.020118343195266272,
+ "grad_norm": 0.6508055329322815,
+ "learning_rate": 0.00011675675675675676,
+ "loss": 0.19072499871253967,
+ "num_input_tokens_seen": 80144,
+ "step": 85,
+ "train_runtime": 334.2158,
+ "train_tokens_per_second": 239.797
+ },
+ {
+ "epoch": 0.020355029585798816,
+ "grad_norm": 0.606562614440918,
+ "learning_rate": 0.00011567567567567568,
+ "loss": 0.20951862633228302,
+ "num_input_tokens_seen": 81108,
+ "step": 86,
+ "train_runtime": 337.3105,
+ "train_tokens_per_second": 240.455
+ },
+ {
+ "epoch": 0.02059171597633136,
+ "grad_norm": 0.6757521629333496,
+ "learning_rate": 0.00011459459459459461,
+ "loss": 0.17614495754241943,
+ "num_input_tokens_seen": 82032,
+ "step": 87,
+ "train_runtime": 340.3438,
+ "train_tokens_per_second": 241.027
+ },
+ {
+ "epoch": 0.020828402366863907,
+ "grad_norm": 0.671481728553772,
+ "learning_rate": 0.00011351351351351351,
+ "loss": 0.1901710331439972,
+ "num_input_tokens_seen": 82954,
+ "step": 88,
+ "train_runtime": 343.4059,
+ "train_tokens_per_second": 241.563
+ },
+ {
+ "epoch": 0.02106508875739645,
+ "grad_norm": 0.5919027924537659,
+ "learning_rate": 0.00011243243243243244,
+ "loss": 0.1563737988471985,
+ "num_input_tokens_seen": 83880,
+ "step": 89,
+ "train_runtime": 346.5125,
+ "train_tokens_per_second": 242.069
+ },
+ {
+ "epoch": 0.021301775147928994,
+ "grad_norm": 0.7108309268951416,
+ "learning_rate": 0.00011135135135135135,
+ "loss": 0.17256894707679749,
+ "num_input_tokens_seen": 84852,
+ "step": 90,
+ "train_runtime": 349.6508,
+ "train_tokens_per_second": 242.676
+ },
+ {
+ "epoch": 0.021538461538461538,
+ "grad_norm": 0.8608420491218567,
+ "learning_rate": 0.00011027027027027029,
+ "loss": 0.1845530867576599,
+ "num_input_tokens_seen": 85820,
+ "step": 91,
+ "train_runtime": 356.267,
+ "train_tokens_per_second": 240.887
+ },
+ {
+ "epoch": 0.02177514792899408,
+ "grad_norm": 0.9677010774612427,
+ "learning_rate": 0.00010918918918918919,
+ "loss": 0.24358105659484863,
+ "num_input_tokens_seen": 86770,
+ "step": 92,
+ "train_runtime": 359.3722,
+ "train_tokens_per_second": 241.449
+ },
+ {
+ "epoch": 0.02201183431952663,
+ "grad_norm": 0.6302966475486755,
+ "learning_rate": 0.00010810810810810812,
+ "loss": 0.18155133724212646,
+ "num_input_tokens_seen": 87736,
+ "step": 93,
+ "train_runtime": 362.5874,
+ "train_tokens_per_second": 241.972
+ },
+ {
+ "epoch": 0.022248520710059172,
+ "grad_norm": 0.7171812653541565,
+ "learning_rate": 0.00010702702702702702,
+ "loss": 0.15349116921424866,
+ "num_input_tokens_seen": 88626,
+ "step": 94,
+ "train_runtime": 365.7055,
+ "train_tokens_per_second": 242.343
+ },
+ {
+ "epoch": 0.022485207100591716,
+ "grad_norm": 0.7326887845993042,
+ "learning_rate": 0.00010594594594594595,
+ "loss": 0.18243758380413055,
+ "num_input_tokens_seen": 89548,
+ "step": 95,
+ "train_runtime": 368.799,
+ "train_tokens_per_second": 242.81
+ },
+ {
+ "epoch": 0.02272189349112426,
+ "grad_norm": 0.6686019897460938,
+ "learning_rate": 0.00010486486486486487,
+ "loss": 0.15622465312480927,
+ "num_input_tokens_seen": 90504,
+ "step": 96,
+ "train_runtime": 371.8404,
+ "train_tokens_per_second": 243.395
+ },
+ {
+ "epoch": 0.022958579881656806,
+ "grad_norm": 0.6306503415107727,
+ "learning_rate": 0.0001037837837837838,
+ "loss": 0.1272117793560028,
+ "num_input_tokens_seen": 91454,
+ "step": 97,
+ "train_runtime": 375.0199,
+ "train_tokens_per_second": 243.864
+ },
+ {
+ "epoch": 0.02319526627218935,
+ "grad_norm": 0.6068493127822876,
+ "learning_rate": 0.0001027027027027027,
+ "loss": 0.1577155441045761,
+ "num_input_tokens_seen": 92466,
+ "step": 98,
+ "train_runtime": 378.0978,
+ "train_tokens_per_second": 244.556
+ },
+ {
+ "epoch": 0.023431952662721894,
+ "grad_norm": 0.7263635993003845,
+ "learning_rate": 0.00010162162162162163,
+ "loss": 0.1845340132713318,
+ "num_input_tokens_seen": 93398,
+ "step": 99,
+ "train_runtime": 381.1358,
+ "train_tokens_per_second": 245.052
+ },
+ {
+ "epoch": 0.023668639053254437,
+ "grad_norm": 0.6466794013977051,
+ "learning_rate": 0.00010054054054054053,
+ "loss": 0.14443959295749664,
+ "num_input_tokens_seen": 94230,
+ "step": 100,
+ "train_runtime": 384.2906,
+ "train_tokens_per_second": 245.205
+ },
+ {
+ "epoch": 0.02390532544378698,
+ "grad_norm": 0.6924043297767639,
+ "learning_rate": 9.945945945945948e-05,
+ "loss": 0.18473166227340698,
+ "num_input_tokens_seen": 95132,
+ "step": 101,
+ "train_runtime": 387.5005,
+ "train_tokens_per_second": 245.502
+ },
+ {
+ "epoch": 0.024142011834319528,
+ "grad_norm": 0.732337236404419,
+ "learning_rate": 9.837837837837839e-05,
+ "loss": 0.16353662312030792,
+ "num_input_tokens_seen": 96080,
+ "step": 102,
+ "train_runtime": 390.584,
+ "train_tokens_per_second": 245.991
+ },
+ {
+ "epoch": 0.02437869822485207,
+ "grad_norm": 0.5956932902336121,
+ "learning_rate": 9.729729729729731e-05,
+ "loss": 0.16359847784042358,
+ "num_input_tokens_seen": 97048,
+ "step": 103,
+ "train_runtime": 393.6474,
+ "train_tokens_per_second": 246.535
+ },
+ {
+ "epoch": 0.024615384615384615,
+ "grad_norm": 0.6137388348579407,
+ "learning_rate": 9.621621621621622e-05,
+ "loss": 0.15779417753219604,
+ "num_input_tokens_seen": 97986,
+ "step": 104,
+ "train_runtime": 396.7042,
+ "train_tokens_per_second": 247.0
+ },
+ {
+ "epoch": 0.02485207100591716,
+ "grad_norm": 0.7936599254608154,
+ "learning_rate": 9.513513513513514e-05,
+ "loss": 0.21357379853725433,
+ "num_input_tokens_seen": 98894,
+ "step": 105,
+ "train_runtime": 399.8228,
+ "train_tokens_per_second": 247.345
+ },
+ {
+ "epoch": 0.025088757396449703,
+ "grad_norm": 0.6574507355690002,
+ "learning_rate": 9.405405405405407e-05,
+ "loss": 0.1685573309659958,
+ "num_input_tokens_seen": 99856,
+ "step": 106,
+ "train_runtime": 402.8867,
+ "train_tokens_per_second": 247.851
+ },
+ {
+ "epoch": 0.02532544378698225,
+ "grad_norm": 0.5891634821891785,
+ "learning_rate": 9.297297297297299e-05,
+ "loss": 0.12082140147686005,
+ "num_input_tokens_seen": 100810,
+ "step": 107,
+ "train_runtime": 405.9752,
+ "train_tokens_per_second": 248.316
+ },
+ {
+ "epoch": 0.025562130177514793,
+ "grad_norm": 0.6624342799186707,
+ "learning_rate": 9.18918918918919e-05,
+ "loss": 0.1473795771598816,
+ "num_input_tokens_seen": 101744,
+ "step": 108,
+ "train_runtime": 409.0084,
+ "train_tokens_per_second": 248.758
+ },
+ {
+ "epoch": 0.025798816568047337,
+ "grad_norm": 0.5968036651611328,
+ "learning_rate": 9.081081081081082e-05,
+ "loss": 0.12923751771450043,
+ "num_input_tokens_seen": 102702,
+ "step": 109,
+ "train_runtime": 412.199,
+ "train_tokens_per_second": 249.156
+ },
+ {
+ "epoch": 0.02603550295857988,
+ "grad_norm": 0.5954580307006836,
+ "learning_rate": 8.972972972972973e-05,
+ "loss": 0.1161399856209755,
+ "num_input_tokens_seen": 103672,
+ "step": 110,
+ "train_runtime": 415.3736,
+ "train_tokens_per_second": 249.587
+ },
+ {
+ "epoch": 0.026272189349112424,
+ "grad_norm": 0.6139379739761353,
+ "learning_rate": 8.864864864864866e-05,
+ "loss": 0.12352899461984634,
+ "num_input_tokens_seen": 104598,
+ "step": 111,
+ "train_runtime": 418.3271,
+ "train_tokens_per_second": 250.039
+ },
+ {
+ "epoch": 0.02650887573964497,
+ "grad_norm": 0.7933095097541809,
+ "learning_rate": 8.756756756756758e-05,
+ "loss": 0.1709994524717331,
+ "num_input_tokens_seen": 105580,
+ "step": 112,
+ "train_runtime": 421.3565,
+ "train_tokens_per_second": 250.572
+ },
+ {
+ "epoch": 0.026745562130177515,
+ "grad_norm": 0.7641370892524719,
+ "learning_rate": 8.64864864864865e-05,
+ "loss": 0.13086311519145966,
+ "num_input_tokens_seen": 106444,
+ "step": 113,
+ "train_runtime": 424.4668,
+ "train_tokens_per_second": 250.771
+ },
+ {
+ "epoch": 0.02698224852071006,
+ "grad_norm": 0.7611938118934631,
+ "learning_rate": 8.540540540540541e-05,
+ "loss": 0.16537640988826752,
+ "num_input_tokens_seen": 107454,
+ "step": 114,
+ "train_runtime": 427.5,
+ "train_tokens_per_second": 251.354
+ },
+ {
+ "epoch": 0.027218934911242602,
+ "grad_norm": 0.7495304346084595,
+ "learning_rate": 8.432432432432433e-05,
+ "loss": 0.14228317141532898,
+ "num_input_tokens_seen": 108426,
+ "step": 115,
+ "train_runtime": 430.5035,
+ "train_tokens_per_second": 251.859
+ },
+ {
+ "epoch": 0.02745562130177515,
+ "grad_norm": 1.2078282833099365,
+ "learning_rate": 8.324324324324326e-05,
+ "loss": 0.10530930012464523,
+ "num_input_tokens_seen": 109446,
+ "step": 116,
+ "train_runtime": 433.4964,
+ "train_tokens_per_second": 252.473
+ },
+ {
+ "epoch": 0.027692307692307693,
+ "grad_norm": 0.6765572428703308,
+ "learning_rate": 8.216216216216217e-05,
+ "loss": 0.10717278718948364,
+ "num_input_tokens_seen": 110374,
+ "step": 117,
+ "train_runtime": 436.5901,
+ "train_tokens_per_second": 252.809
+ },
+ {
+ "epoch": 0.027928994082840237,
+ "grad_norm": 0.8501551151275635,
+ "learning_rate": 8.108108108108109e-05,
+ "loss": 0.17784273624420166,
+ "num_input_tokens_seen": 111340,
+ "step": 118,
+ "train_runtime": 439.6526,
+ "train_tokens_per_second": 253.245
+ },
+ {
+ "epoch": 0.02816568047337278,
+ "grad_norm": 0.7583926320075989,
+ "learning_rate": 8e-05,
+ "loss": 0.16588062047958374,
+ "num_input_tokens_seen": 112288,
+ "step": 119,
+ "train_runtime": 442.8046,
+ "train_tokens_per_second": 253.584
+ },
+ {
+ "epoch": 0.028402366863905324,
+ "grad_norm": 0.6620173454284668,
+ "learning_rate": 7.891891891891892e-05,
+ "loss": 0.11676295101642609,
+ "num_input_tokens_seen": 113206,
+ "step": 120,
+ "train_runtime": 445.9682,
+ "train_tokens_per_second": 253.843
+ }
+ ],
+ "logging_steps": 1,
+ "max_steps": 190,
+ "num_input_tokens_seen": 113206,
+ "num_train_epochs": 1,
+ "save_steps": 30,
+ "stateful_callbacks": {
+ "TrainerControl": {
+ "args": {
+ "should_epoch_stop": false,
+ "should_evaluate": false,
+ "should_log": false,
+ "should_save": true,
+ "should_training_stop": false
+ },
+ "attributes": {}
+ }
+ },
+ "total_flos": 1632037070282880.0,
+ "train_batch_size": 2,
+ "trial_name": null,
+ "trial_params": null
+}
diff --git a/checkpoint-120/training_args.bin b/checkpoint-120/training_args.bin
new file mode 100644
index 0000000000000000000000000000000000000000..cefffe2bb567bdf18003e5cc5aaf746472c2981f
--- /dev/null
+++ b/checkpoint-120/training_args.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:34c4ca6d50f04301fb38588516a91421e14a8ad1ffddf12afdfd9ba35cd4b64e
+size 5777
diff --git a/checkpoint-150/README.md b/checkpoint-150/README.md
new file mode 100644
index 0000000000000000000000000000000000000000..41470ec6557bbc37155893f70bd59986401b9148
--- /dev/null
+++ b/checkpoint-150/README.md
@@ -0,0 +1,210 @@
+---
+base_model: unsloth/gemma-4-e2b-it-unsloth-bnb-4bit
+library_name: peft
+pipeline_tag: text-generation
+tags:
+- base_model:adapter:unsloth/gemma-4-e2b-it-unsloth-bnb-4bit
+- lora
+- sft
+- transformers
+- trl
+- unsloth
+---
+
+# Model Card for Model ID
+
+
+
+
+
+## Model Details
+
+### Model Description
+
+
+
+
+
+- **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]
+
+
+
+- **Repository:** [More Information Needed]
+- **Paper [optional]:** [More Information Needed]
+- **Demo [optional]:** [More Information Needed]
+
+## Uses
+
+
+
+### Direct Use
+
+
+
+[More Information Needed]
+
+### Downstream Use [optional]
+
+
+
+[More Information Needed]
+
+### Out-of-Scope Use
+
+
+
+[More Information Needed]
+
+## Bias, Risks, and Limitations
+
+
+
+[More Information Needed]
+
+### Recommendations
+
+
+
+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
+
+
+
+[More Information Needed]
+
+### Training Procedure
+
+
+
+#### Preprocessing [optional]
+
+[More Information Needed]
+
+
+#### Training Hyperparameters
+
+- **Training regime:** [More Information Needed]
+
+#### Speeds, Sizes, Times [optional]
+
+
+
+[More Information Needed]
+
+## Evaluation
+
+
+
+### Testing Data, Factors & Metrics
+
+#### Testing Data
+
+
+
+[More Information Needed]
+
+#### Factors
+
+
+
+[More Information Needed]
+
+#### Metrics
+
+
+
+[More Information Needed]
+
+### Results
+
+[More Information Needed]
+
+#### Summary
+
+
+
+## Model Examination [optional]
+
+
+
+[More Information Needed]
+
+## Environmental Impact
+
+
+
+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]
+
+
+
+**BibTeX:**
+
+[More Information Needed]
+
+**APA:**
+
+[More Information Needed]
+
+## Glossary [optional]
+
+
+
+[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
\ No newline at end of file
diff --git a/checkpoint-150/adapter_config.json b/checkpoint-150/adapter_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..915726be1cd67eb165724a8390b51a57fe0dd6df
--- /dev/null
+++ b/checkpoint-150/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": "unsloth/gemma-4-e2b-it-unsloth-bnb-4bit",
+ "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": [
+ "q_proj",
+ "gate_proj",
+ "v_proj",
+ "k_proj",
+ "down_proj",
+ "o_proj",
+ "up_proj"
+ ],
+ "target_parameters": null,
+ "task_type": "CAUSAL_LM",
+ "trainable_token_indices": null,
+ "use_dora": false,
+ "use_qalora": false,
+ "use_rslora": false
+}
\ No newline at end of file
diff --git a/checkpoint-150/adapter_model.safetensors b/checkpoint-150/adapter_model.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..c787305432866f090776eef524af0ac379f6aaad
--- /dev/null
+++ b/checkpoint-150/adapter_model.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:79025ee3acf1d35d8aa435821bda964bc6b3cb6801665ea77bdff85aaa1067c2
+size 124277728
diff --git a/checkpoint-150/chat_template.jinja b/checkpoint-150/chat_template.jinja
new file mode 100644
index 0000000000000000000000000000000000000000..dc032e46e28e0d325b18d26da1279cdda3afa8d7
--- /dev/null
+++ b/checkpoint-150/chat_template.jinja
@@ -0,0 +1,351 @@
+{%- macro format_parameters(properties, required, filter_keys=false) -%}
+ {%- 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 not filter_keys or 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([]), filter_keys=true) -}}
+ }
+ {%- 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('') -%}
+ {%- 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 -%}
+ {{- '' -}}
+{%- 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'] -%}
+ {%- if messages[0]['content'] is string -%}
+ {{- messages[0]['content'] | trim -}}
+ {%- elif messages[0]['content'] is sequence -%}
+ {%- for item in messages[0]['content'] -%}
+ {{- item['text'] | trim + ' '-}}
+ {%- endfor -%}
+ {%- endif -%}
+ {%- set loop_messages = messages[1:] -%}
+ {%- endif -%}
+ {%- if tools -%}
+ {%- for tool in tools %}
+ {{- '<|tool>' -}}
+ {{- format_function_declaration(tool) | trim -}}
+ {{- '' -}}
+ {%- endfor %}
+ {%- set ns.prev_message_type = 'tool' -%}
+ {%- endif -%}
+ {{- '\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' -}}
+ {%- 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 -%}
+ {{- '}' -}}
+ {%- 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 -%}
+
+ {%- set captured_content -%}
+ {%- 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 -%}
+ {%- endset -%}
+
+ {{- captured_content -}}
+ {%- set has_content = captured_content | trim | length > 0 -%}
+
+ {%- if ns.prev_message_type == 'tool_call' and not ns_tr_out.flag -%}
+ {{- '<|tool_response>' -}}
+ {%- elif not (ns_tr_out.flag and not has_content) -%}
+ {{- '\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 -%}
\ No newline at end of file
diff --git a/checkpoint-150/optimizer.pt b/checkpoint-150/optimizer.pt
new file mode 100644
index 0000000000000000000000000000000000000000..ab8aa79fe1b63cff8272e54e6ea0f6e0f35e1cd2
--- /dev/null
+++ b/checkpoint-150/optimizer.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:676e7ff5f1af2f689ece698e6985cdaf684ab2bafbeeaa8681977a6453dffe22
+size 52047117
diff --git a/checkpoint-150/processor_config.json b/checkpoint-150/processor_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..bfaa5e360ad17c2761b7ee3b05abf03e381e8aa1
--- /dev/null
+++ b/checkpoint-150/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"
+ }
+}
diff --git a/checkpoint-150/rng_state.pth b/checkpoint-150/rng_state.pth
new file mode 100644
index 0000000000000000000000000000000000000000..3ef66339b9befa098183fd5d69faed6838e526b0
--- /dev/null
+++ b/checkpoint-150/rng_state.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f1d565802a8e26c4e8a31328752b7a7fdc186d9401aa008e65697d0ad8c22e33
+size 14645
diff --git a/checkpoint-150/scaler.pt b/checkpoint-150/scaler.pt
new file mode 100644
index 0000000000000000000000000000000000000000..5e3ce5fc7ee2fcd8a46c258277b182352c1b6696
--- /dev/null
+++ b/checkpoint-150/scaler.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:41ce6f17608f83ea4a19655ee2f08447977bad8cf4b831826d8c03f86b5176e9
+size 1383
diff --git a/checkpoint-150/scheduler.pt b/checkpoint-150/scheduler.pt
new file mode 100644
index 0000000000000000000000000000000000000000..91371002aba314039408a1adad72709bea076d7b
--- /dev/null
+++ b/checkpoint-150/scheduler.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a02bbb7a0dfd9017bf6ee57d7f39ecba548a46e11e466823a59cfab4c980712f
+size 1465
diff --git a/checkpoint-150/tokenizer.json b/checkpoint-150/tokenizer.json
new file mode 100644
index 0000000000000000000000000000000000000000..1ff9f3e3439a939b971f9919e821bf87e835a503
--- /dev/null
+++ b/checkpoint-150/tokenizer.json
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:cc8d3a0ce36466ccc1278bf987df5f71db1719b9ca6b4118264f45cb627bfe0f
+size 32169626
diff --git a/checkpoint-150/tokenizer_config.json b/checkpoint-150/tokenizer_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..24b1522bb32b0eb05b631b1a9f112751b8cb0fab
--- /dev/null
+++ b/checkpoint-150/tokenizer_config.json
@@ -0,0 +1,289 @@
+{
+ "audio_token": "<|audio|>",
+ "backend": "tokenizers",
+ "boa_token": "<|audio>",
+ "boi_token": "<|image>",
+ "bos_token": "",
+ "eoa_token": "",
+ "eoc_token": "",
+ "eoi_token": "",
+ "eos_token": "",
+ "eot_token": "",
+ "escape_token": "<|\"|>",
+ "etc_token": "",
+ "etd_token": "",
+ "etr_token": "",
+ "extra_special_tokens": [
+ "<|video|>"
+ ],
+ "image_token": "<|image|>",
+ "is_local": false,
+ "mask_token": "",
+ "model_max_length": 131072,
+ "model_specific_special_tokens": {
+ "audio_token": "<|audio|>",
+ "boa_token": "<|audio>",
+ "boi_token": "<|image>",
+ "eoa_token": "",
+ "eoc_token": "",
+ "eoi_token": "",
+ "eot_token": "",
+ "escape_token": "<|\"|>",
+ "etc_token": "",
+ "etd_token": "",
+ "etr_token": "",
+ "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": "",
+ "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\\w+)(?P\\{.*\\})"
+ },
+ "type": {
+ "const": "function"
+ }
+ },
+ "type": "object"
+ },
+ "type": "array",
+ "x-regex-iterator": "<\\|tool_call>(.*?)"
+ }
+ },
+ "type": "object",
+ "x-regex": "(\\<\\|channel\\>thought\\n(?P.*?)\\)?(?P\\<\\|tool_call\\>.*\\)?(?P(?:(?!\\)(?!\\<\\|tool_response\\>).)+)?(?:\\|\\<\\|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": "",
+ "added_tokens_decoder": {
+ "0": {
+ "content": "",
+ "single_word": false,
+ "lstrip": false,
+ "rstrip": false,
+ "normalized": false,
+ "special": true
+ },
+ "1": {
+ "content": "",
+ "single_word": false,
+ "lstrip": false,
+ "rstrip": false,
+ "normalized": false,
+ "special": true
+ },
+ "2": {
+ "content": "",
+ "single_word": false,
+ "lstrip": false,
+ "rstrip": false,
+ "normalized": false,
+ "special": true
+ },
+ "3": {
+ "content": "",
+ "single_word": false,
+ "lstrip": false,
+ "rstrip": false,
+ "normalized": false,
+ "special": true
+ },
+ "4": {
+ "content": "",
+ "single_word": false,
+ "lstrip": false,
+ "rstrip": false,
+ "normalized": false,
+ "special": true
+ },
+ "46": {
+ "content": "<|tool>",
+ "single_word": false,
+ "lstrip": false,
+ "rstrip": false,
+ "normalized": false,
+ "special": true
+ },
+ "47": {
+ "content": "",
+ "single_word": false,
+ "lstrip": false,
+ "rstrip": false,
+ "normalized": false,
+ "special": true
+ },
+ "48": {
+ "content": "<|tool_call>",
+ "single_word": false,
+ "lstrip": false,
+ "rstrip": false,
+ "normalized": false,
+ "special": true
+ },
+ "49": {
+ "content": "",
+ "single_word": false,
+ "lstrip": false,
+ "rstrip": false,
+ "normalized": false,
+ "special": true
+ },
+ "50": {
+ "content": "<|tool_response>",
+ "single_word": false,
+ "lstrip": false,
+ "rstrip": false,
+ "normalized": false,
+ "special": true
+ },
+ "51": {
+ "content": "",
+ "single_word": false,
+ "lstrip": false,
+ "rstrip": false,
+ "normalized": false,
+ "special": true
+ },
+ "52": {
+ "content": "<|\"|>",
+ "single_word": false,
+ "lstrip": false,
+ "rstrip": false,
+ "normalized": false,
+ "special": true
+ },
+ "98": {
+ "content": "<|think|>",
+ "single_word": false,
+ "lstrip": false,
+ "rstrip": false,
+ "normalized": false,
+ "special": true
+ },
+ "100": {
+ "content": "<|channel>",
+ "single_word": false,
+ "lstrip": false,
+ "rstrip": false,
+ "normalized": false,
+ "special": true
+ },
+ "101": {
+ "content": "",
+ "single_word": false,
+ "lstrip": false,
+ "rstrip": false,
+ "normalized": false,
+ "special": true
+ },
+ "105": {
+ "content": "<|turn>",
+ "single_word": false,
+ "lstrip": false,
+ "rstrip": false,
+ "normalized": false,
+ "special": true
+ },
+ "106": {
+ "content": "",
+ "single_word": false,
+ "lstrip": false,
+ "rstrip": false,
+ "normalized": false,
+ "special": true
+ },
+ "255999": {
+ "content": "<|image>",
+ "single_word": false,
+ "lstrip": false,
+ "rstrip": false,
+ "normalized": false,
+ "special": true
+ },
+ "256000": {
+ "content": "<|audio>",
+ "single_word": false,
+ "lstrip": false,
+ "rstrip": false,
+ "normalized": false,
+ "special": true
+ },
+ "258880": {
+ "content": "<|image|>",
+ "single_word": false,
+ "lstrip": false,
+ "rstrip": false,
+ "normalized": false,
+ "special": true
+ },
+ "258881": {
+ "content": "<|audio|>",
+ "single_word": false,
+ "lstrip": false,
+ "rstrip": false,
+ "normalized": false,
+ "special": true
+ },
+ "258882": {
+ "content": "",
+ "single_word": false,
+ "lstrip": false,
+ "rstrip": false,
+ "normalized": false,
+ "special": true
+ },
+ "258883": {
+ "content": "",
+ "single_word": false,
+ "lstrip": false,
+ "rstrip": false,
+ "normalized": false,
+ "special": true
+ },
+ "258884": {
+ "content": "<|video|>",
+ "single_word": false,
+ "lstrip": false,
+ "rstrip": false,
+ "normalized": false,
+ "special": true
+ }
+ }
+}
diff --git a/checkpoint-150/trainer_state.json b/checkpoint-150/trainer_state.json
new file mode 100644
index 0000000000000000000000000000000000000000..9c7c1589da8ffa2653b1a405485468e519b1a768
--- /dev/null
+++ b/checkpoint-150/trainer_state.json
@@ -0,0 +1,1534 @@
+{
+ "best_global_step": null,
+ "best_metric": null,
+ "best_model_checkpoint": null,
+ "epoch": 0.03550295857988166,
+ "eval_steps": 500,
+ "global_step": 150,
+ "is_hyper_param_search": false,
+ "is_local_process_zero": true,
+ "is_world_process_zero": true,
+ "log_history": [
+ {
+ "epoch": 0.00023668639053254438,
+ "grad_norm": NaN,
+ "learning_rate": 0.0,
+ "loss": 4.347962379455566,
+ "num_input_tokens_seen": 842,
+ "step": 1,
+ "train_runtime": 63.0289,
+ "train_tokens_per_second": 13.359
+ },
+ {
+ "epoch": 0.00047337278106508875,
+ "grad_norm": NaN,
+ "learning_rate": 0.0,
+ "loss": 4.634688854217529,
+ "num_input_tokens_seen": 1820,
+ "step": 2,
+ "train_runtime": 67.7033,
+ "train_tokens_per_second": 26.882
+ },
+ {
+ "epoch": 0.0007100591715976331,
+ "grad_norm": 13.944212913513184,
+ "learning_rate": 0.0,
+ "loss": 4.42241096496582,
+ "num_input_tokens_seen": 2754,
+ "step": 3,
+ "train_runtime": 72.324,
+ "train_tokens_per_second": 38.079
+ },
+ {
+ "epoch": 0.0009467455621301775,
+ "grad_norm": 13.445413589477539,
+ "learning_rate": 4e-05,
+ "loss": 4.682804107666016,
+ "num_input_tokens_seen": 3700,
+ "step": 4,
+ "train_runtime": 75.3925,
+ "train_tokens_per_second": 49.077
+ },
+ {
+ "epoch": 0.001183431952662722,
+ "grad_norm": 10.576727867126465,
+ "learning_rate": 8e-05,
+ "loss": 4.166159629821777,
+ "num_input_tokens_seen": 4714,
+ "step": 5,
+ "train_runtime": 78.474,
+ "train_tokens_per_second": 60.071
+ },
+ {
+ "epoch": 0.0014201183431952662,
+ "grad_norm": 8.077850341796875,
+ "learning_rate": 0.00012,
+ "loss": 3.2006261348724365,
+ "num_input_tokens_seen": 5674,
+ "step": 6,
+ "train_runtime": 81.6139,
+ "train_tokens_per_second": 69.522
+ },
+ {
+ "epoch": 0.0016568047337278107,
+ "grad_norm": 6.60789680480957,
+ "learning_rate": 0.00016,
+ "loss": 2.7067415714263916,
+ "num_input_tokens_seen": 6646,
+ "step": 7,
+ "train_runtime": 84.786,
+ "train_tokens_per_second": 78.386
+ },
+ {
+ "epoch": 0.001893491124260355,
+ "grad_norm": 9.175209045410156,
+ "learning_rate": 0.0002,
+ "loss": 2.367570638656616,
+ "num_input_tokens_seen": 7518,
+ "step": 8,
+ "train_runtime": 87.8178,
+ "train_tokens_per_second": 85.609
+ },
+ {
+ "epoch": 0.0021301775147928993,
+ "grad_norm": 3.438948631286621,
+ "learning_rate": 0.00019891891891891895,
+ "loss": 2.1013119220733643,
+ "num_input_tokens_seen": 8470,
+ "step": 9,
+ "train_runtime": 90.9458,
+ "train_tokens_per_second": 93.132
+ },
+ {
+ "epoch": 0.002366863905325444,
+ "grad_norm": 3.459472894668579,
+ "learning_rate": 0.00019783783783783784,
+ "loss": 1.8991940021514893,
+ "num_input_tokens_seen": 9458,
+ "step": 10,
+ "train_runtime": 94.0577,
+ "train_tokens_per_second": 100.555
+ },
+ {
+ "epoch": 0.0026035502958579883,
+ "grad_norm": 2.169750928878784,
+ "learning_rate": 0.00019675675675675678,
+ "loss": 1.7395055294036865,
+ "num_input_tokens_seen": 10456,
+ "step": 11,
+ "train_runtime": 97.2068,
+ "train_tokens_per_second": 107.565
+ },
+ {
+ "epoch": 0.0028402366863905324,
+ "grad_norm": 1.8638927936553955,
+ "learning_rate": 0.00019567567567567567,
+ "loss": 1.5622950792312622,
+ "num_input_tokens_seen": 11410,
+ "step": 12,
+ "train_runtime": 100.2911,
+ "train_tokens_per_second": 113.769
+ },
+ {
+ "epoch": 0.003076923076923077,
+ "grad_norm": 1.4741116762161255,
+ "learning_rate": 0.00019459459459459462,
+ "loss": 1.392699956893921,
+ "num_input_tokens_seen": 12316,
+ "step": 13,
+ "train_runtime": 103.3664,
+ "train_tokens_per_second": 119.149
+ },
+ {
+ "epoch": 0.0033136094674556214,
+ "grad_norm": 1.3238699436187744,
+ "learning_rate": 0.00019351351351351353,
+ "loss": 1.2535074949264526,
+ "num_input_tokens_seen": 13256,
+ "step": 14,
+ "train_runtime": 106.4031,
+ "train_tokens_per_second": 124.583
+ },
+ {
+ "epoch": 0.0035502958579881655,
+ "grad_norm": 1.276484489440918,
+ "learning_rate": 0.00019243243243243245,
+ "loss": 1.135048508644104,
+ "num_input_tokens_seen": 14228,
+ "step": 15,
+ "train_runtime": 109.4909,
+ "train_tokens_per_second": 129.947
+ },
+ {
+ "epoch": 0.00378698224852071,
+ "grad_norm": 1.1560865640640259,
+ "learning_rate": 0.00019135135135135137,
+ "loss": 1.0317097902297974,
+ "num_input_tokens_seen": 15192,
+ "step": 16,
+ "train_runtime": 112.5741,
+ "train_tokens_per_second": 134.951
+ },
+ {
+ "epoch": 0.004023668639053254,
+ "grad_norm": 1.1331353187561035,
+ "learning_rate": 0.00019027027027027028,
+ "loss": 0.9967219233512878,
+ "num_input_tokens_seen": 16130,
+ "step": 17,
+ "train_runtime": 115.6495,
+ "train_tokens_per_second": 139.473
+ },
+ {
+ "epoch": 0.004260355029585799,
+ "grad_norm": 1.1567702293395996,
+ "learning_rate": 0.0001891891891891892,
+ "loss": 0.905368447303772,
+ "num_input_tokens_seen": 17078,
+ "step": 18,
+ "train_runtime": 118.6796,
+ "train_tokens_per_second": 143.9
+ },
+ {
+ "epoch": 0.004497041420118343,
+ "grad_norm": 0.9157600998878479,
+ "learning_rate": 0.00018810810810810814,
+ "loss": 0.8075690865516663,
+ "num_input_tokens_seen": 18062,
+ "step": 19,
+ "train_runtime": 121.8602,
+ "train_tokens_per_second": 148.219
+ },
+ {
+ "epoch": 0.004733727810650888,
+ "grad_norm": 0.8306605815887451,
+ "learning_rate": 0.00018702702702702703,
+ "loss": 0.8007175922393799,
+ "num_input_tokens_seen": 19078,
+ "step": 20,
+ "train_runtime": 125.0019,
+ "train_tokens_per_second": 152.622
+ },
+ {
+ "epoch": 0.004970414201183432,
+ "grad_norm": 0.690018355846405,
+ "learning_rate": 0.00018594594594594597,
+ "loss": 0.7810791730880737,
+ "num_input_tokens_seen": 20068,
+ "step": 21,
+ "train_runtime": 128.0932,
+ "train_tokens_per_second": 156.667
+ },
+ {
+ "epoch": 0.005207100591715977,
+ "grad_norm": 1.4768867492675781,
+ "learning_rate": 0.00018486486486486486,
+ "loss": 0.7172120809555054,
+ "num_input_tokens_seen": 20972,
+ "step": 22,
+ "train_runtime": 131.0901,
+ "train_tokens_per_second": 159.982
+ },
+ {
+ "epoch": 0.00544378698224852,
+ "grad_norm": 0.7999809980392456,
+ "learning_rate": 0.0001837837837837838,
+ "loss": 0.6333726048469543,
+ "num_input_tokens_seen": 21900,
+ "step": 23,
+ "train_runtime": 134.212,
+ "train_tokens_per_second": 163.175
+ },
+ {
+ "epoch": 0.005680473372781065,
+ "grad_norm": 0.9285996556282043,
+ "learning_rate": 0.00018270270270270272,
+ "loss": 0.6750729084014893,
+ "num_input_tokens_seen": 22770,
+ "step": 24,
+ "train_runtime": 137.3193,
+ "train_tokens_per_second": 165.818
+ },
+ {
+ "epoch": 0.005917159763313609,
+ "grad_norm": 0.7733820080757141,
+ "learning_rate": 0.00018162162162162164,
+ "loss": 0.672032356262207,
+ "num_input_tokens_seen": 23746,
+ "step": 25,
+ "train_runtime": 140.3175,
+ "train_tokens_per_second": 169.23
+ },
+ {
+ "epoch": 0.006153846153846154,
+ "grad_norm": 0.73944091796875,
+ "learning_rate": 0.00018054054054054055,
+ "loss": 0.6716101169586182,
+ "num_input_tokens_seen": 24716,
+ "step": 26,
+ "train_runtime": 143.3683,
+ "train_tokens_per_second": 172.395
+ },
+ {
+ "epoch": 0.006390532544378698,
+ "grad_norm": 0.7659372091293335,
+ "learning_rate": 0.00017945945945945947,
+ "loss": 0.6067175269126892,
+ "num_input_tokens_seen": 25724,
+ "step": 27,
+ "train_runtime": 146.5377,
+ "train_tokens_per_second": 175.545
+ },
+ {
+ "epoch": 0.006627218934911243,
+ "grad_norm": 0.5428617000579834,
+ "learning_rate": 0.00017837837837837839,
+ "loss": 0.6083395481109619,
+ "num_input_tokens_seen": 26712,
+ "step": 28,
+ "train_runtime": 149.6429,
+ "train_tokens_per_second": 178.505
+ },
+ {
+ "epoch": 0.006863905325443787,
+ "grad_norm": 0.5915114283561707,
+ "learning_rate": 0.00017729729729729733,
+ "loss": 0.6463742256164551,
+ "num_input_tokens_seen": 27704,
+ "step": 29,
+ "train_runtime": 152.7501,
+ "train_tokens_per_second": 181.368
+ },
+ {
+ "epoch": 0.007100591715976331,
+ "grad_norm": 1.3350757360458374,
+ "learning_rate": 0.00017621621621621622,
+ "loss": 0.5705698132514954,
+ "num_input_tokens_seen": 28648,
+ "step": 30,
+ "train_runtime": 155.8918,
+ "train_tokens_per_second": 183.768
+ },
+ {
+ "epoch": 0.0073372781065088755,
+ "grad_norm": 0.68421471118927,
+ "learning_rate": 0.00017513513513513516,
+ "loss": 0.5660746097564697,
+ "num_input_tokens_seen": 29622,
+ "step": 31,
+ "train_runtime": 163.3209,
+ "train_tokens_per_second": 181.373
+ },
+ {
+ "epoch": 0.00757396449704142,
+ "grad_norm": 0.5762604475021362,
+ "learning_rate": 0.00017405405405405405,
+ "loss": 0.5483309030532837,
+ "num_input_tokens_seen": 30534,
+ "step": 32,
+ "train_runtime": 166.5185,
+ "train_tokens_per_second": 183.367
+ },
+ {
+ "epoch": 0.0078106508875739646,
+ "grad_norm": 0.5751784443855286,
+ "learning_rate": 0.000172972972972973,
+ "loss": 0.5207005739212036,
+ "num_input_tokens_seen": 31458,
+ "step": 33,
+ "train_runtime": 169.7055,
+ "train_tokens_per_second": 185.368
+ },
+ {
+ "epoch": 0.008047337278106508,
+ "grad_norm": 0.5752039551734924,
+ "learning_rate": 0.0001718918918918919,
+ "loss": 0.5534584522247314,
+ "num_input_tokens_seen": 32424,
+ "step": 34,
+ "train_runtime": 172.9184,
+ "train_tokens_per_second": 187.51
+ },
+ {
+ "epoch": 0.008284023668639054,
+ "grad_norm": 0.5918363332748413,
+ "learning_rate": 0.00017081081081081083,
+ "loss": 0.482683002948761,
+ "num_input_tokens_seen": 33398,
+ "step": 35,
+ "train_runtime": 176.2158,
+ "train_tokens_per_second": 189.529
+ },
+ {
+ "epoch": 0.008520710059171597,
+ "grad_norm": 0.6921846270561218,
+ "learning_rate": 0.00016972972972972974,
+ "loss": 0.46112507581710815,
+ "num_input_tokens_seen": 34324,
+ "step": 36,
+ "train_runtime": 179.3173,
+ "train_tokens_per_second": 191.415
+ },
+ {
+ "epoch": 0.008757396449704143,
+ "grad_norm": 0.6429721117019653,
+ "learning_rate": 0.00016864864864864866,
+ "loss": 0.45629197359085083,
+ "num_input_tokens_seen": 35270,
+ "step": 37,
+ "train_runtime": 182.3558,
+ "train_tokens_per_second": 193.413
+ },
+ {
+ "epoch": 0.008994082840236686,
+ "grad_norm": 0.5477020740509033,
+ "learning_rate": 0.00016756756756756757,
+ "loss": 0.4547877907752991,
+ "num_input_tokens_seen": 36206,
+ "step": 38,
+ "train_runtime": 185.4483,
+ "train_tokens_per_second": 195.235
+ },
+ {
+ "epoch": 0.009230769230769232,
+ "grad_norm": 0.6158634424209595,
+ "learning_rate": 0.00016648648648648652,
+ "loss": 0.367482990026474,
+ "num_input_tokens_seen": 37080,
+ "step": 39,
+ "train_runtime": 188.5114,
+ "train_tokens_per_second": 196.699
+ },
+ {
+ "epoch": 0.009467455621301775,
+ "grad_norm": 0.5405091643333435,
+ "learning_rate": 0.0001654054054054054,
+ "loss": 0.42049410939216614,
+ "num_input_tokens_seen": 38030,
+ "step": 40,
+ "train_runtime": 191.5986,
+ "train_tokens_per_second": 198.488
+ },
+ {
+ "epoch": 0.009704142011834319,
+ "grad_norm": 0.6389183402061462,
+ "learning_rate": 0.00016432432432432435,
+ "loss": 0.3935014307498932,
+ "num_input_tokens_seen": 38942,
+ "step": 41,
+ "train_runtime": 194.8426,
+ "train_tokens_per_second": 199.864
+ },
+ {
+ "epoch": 0.009940828402366864,
+ "grad_norm": 0.6742956042289734,
+ "learning_rate": 0.00016324324324324324,
+ "loss": 0.4373692274093628,
+ "num_input_tokens_seen": 39902,
+ "step": 42,
+ "train_runtime": 197.8938,
+ "train_tokens_per_second": 201.633
+ },
+ {
+ "epoch": 0.010177514792899408,
+ "grad_norm": 0.6083794236183167,
+ "learning_rate": 0.00016216216216216218,
+ "loss": 0.3776765465736389,
+ "num_input_tokens_seen": 40858,
+ "step": 43,
+ "train_runtime": 200.9708,
+ "train_tokens_per_second": 203.303
+ },
+ {
+ "epoch": 0.010414201183431953,
+ "grad_norm": 0.5469992160797119,
+ "learning_rate": 0.0001610810810810811,
+ "loss": 0.31133151054382324,
+ "num_input_tokens_seen": 41760,
+ "step": 44,
+ "train_runtime": 204.1304,
+ "train_tokens_per_second": 204.575
+ },
+ {
+ "epoch": 0.010650887573964497,
+ "grad_norm": 0.6452498435974121,
+ "learning_rate": 0.00016,
+ "loss": 0.42721521854400635,
+ "num_input_tokens_seen": 42740,
+ "step": 45,
+ "train_runtime": 207.1845,
+ "train_tokens_per_second": 206.29
+ },
+ {
+ "epoch": 0.01088757396449704,
+ "grad_norm": 0.5988945960998535,
+ "learning_rate": 0.00015891891891891893,
+ "loss": 0.41375258564949036,
+ "num_input_tokens_seen": 43750,
+ "step": 46,
+ "train_runtime": 210.2912,
+ "train_tokens_per_second": 208.045
+ },
+ {
+ "epoch": 0.011124260355029586,
+ "grad_norm": 0.6452038288116455,
+ "learning_rate": 0.00015783783783783785,
+ "loss": 0.3485276699066162,
+ "num_input_tokens_seen": 44690,
+ "step": 47,
+ "train_runtime": 213.3684,
+ "train_tokens_per_second": 209.45
+ },
+ {
+ "epoch": 0.01136094674556213,
+ "grad_norm": 0.7041454911231995,
+ "learning_rate": 0.00015675675675675676,
+ "loss": 0.3296637535095215,
+ "num_input_tokens_seen": 45666,
+ "step": 48,
+ "train_runtime": 216.4459,
+ "train_tokens_per_second": 210.981
+ },
+ {
+ "epoch": 0.011597633136094675,
+ "grad_norm": 0.6726324558258057,
+ "learning_rate": 0.0001556756756756757,
+ "loss": 0.3179457187652588,
+ "num_input_tokens_seen": 46560,
+ "step": 49,
+ "train_runtime": 219.4697,
+ "train_tokens_per_second": 212.148
+ },
+ {
+ "epoch": 0.011834319526627219,
+ "grad_norm": 0.7513703107833862,
+ "learning_rate": 0.0001545945945945946,
+ "loss": 0.3722391426563263,
+ "num_input_tokens_seen": 47528,
+ "step": 50,
+ "train_runtime": 222.596,
+ "train_tokens_per_second": 213.517
+ },
+ {
+ "epoch": 0.012071005917159764,
+ "grad_norm": 0.6454160213470459,
+ "learning_rate": 0.00015351351351351354,
+ "loss": 0.283376544713974,
+ "num_input_tokens_seen": 48468,
+ "step": 51,
+ "train_runtime": 225.7262,
+ "train_tokens_per_second": 214.72
+ },
+ {
+ "epoch": 0.012307692307692308,
+ "grad_norm": 0.9116244316101074,
+ "learning_rate": 0.00015243243243243243,
+ "loss": 0.4286752939224243,
+ "num_input_tokens_seen": 49444,
+ "step": 52,
+ "train_runtime": 228.803,
+ "train_tokens_per_second": 216.099
+ },
+ {
+ "epoch": 0.012544378698224851,
+ "grad_norm": 0.6157099604606628,
+ "learning_rate": 0.00015135135135135137,
+ "loss": 0.27355721592903137,
+ "num_input_tokens_seen": 50424,
+ "step": 53,
+ "train_runtime": 231.8413,
+ "train_tokens_per_second": 217.494
+ },
+ {
+ "epoch": 0.012781065088757397,
+ "grad_norm": 0.6335808038711548,
+ "learning_rate": 0.00015027027027027028,
+ "loss": 0.37197017669677734,
+ "num_input_tokens_seen": 51328,
+ "step": 54,
+ "train_runtime": 234.9265,
+ "train_tokens_per_second": 218.485
+ },
+ {
+ "epoch": 0.01301775147928994,
+ "grad_norm": 0.6838424205780029,
+ "learning_rate": 0.0001491891891891892,
+ "loss": 0.3504602909088135,
+ "num_input_tokens_seen": 52274,
+ "step": 55,
+ "train_runtime": 238.0114,
+ "train_tokens_per_second": 219.628
+ },
+ {
+ "epoch": 0.013254437869822486,
+ "grad_norm": 0.6566418409347534,
+ "learning_rate": 0.00014810810810810812,
+ "loss": 0.3189927935600281,
+ "num_input_tokens_seen": 53214,
+ "step": 56,
+ "train_runtime": 241.0892,
+ "train_tokens_per_second": 220.723
+ },
+ {
+ "epoch": 0.01349112426035503,
+ "grad_norm": 0.6112971901893616,
+ "learning_rate": 0.00014702702702702703,
+ "loss": 0.3080369830131531,
+ "num_input_tokens_seen": 54162,
+ "step": 57,
+ "train_runtime": 244.2461,
+ "train_tokens_per_second": 221.752
+ },
+ {
+ "epoch": 0.013727810650887575,
+ "grad_norm": 0.6469757556915283,
+ "learning_rate": 0.00014594594594594595,
+ "loss": 0.24279898405075073,
+ "num_input_tokens_seen": 55158,
+ "step": 58,
+ "train_runtime": 247.4064,
+ "train_tokens_per_second": 222.945
+ },
+ {
+ "epoch": 0.013964497041420118,
+ "grad_norm": 0.7099384665489197,
+ "learning_rate": 0.0001448648648648649,
+ "loss": 0.31188705563545227,
+ "num_input_tokens_seen": 56096,
+ "step": 59,
+ "train_runtime": 250.4708,
+ "train_tokens_per_second": 223.962
+ },
+ {
+ "epoch": 0.014201183431952662,
+ "grad_norm": 0.5927040576934814,
+ "learning_rate": 0.00014378378378378378,
+ "loss": 0.25696688890457153,
+ "num_input_tokens_seen": 57012,
+ "step": 60,
+ "train_runtime": 253.7131,
+ "train_tokens_per_second": 224.711
+ },
+ {
+ "epoch": 0.014437869822485207,
+ "grad_norm": 0.5483338236808777,
+ "learning_rate": 0.00014270270270270272,
+ "loss": 0.2725352644920349,
+ "num_input_tokens_seen": 58036,
+ "step": 61,
+ "train_runtime": 260.2844,
+ "train_tokens_per_second": 222.971
+ },
+ {
+ "epoch": 0.014674556213017751,
+ "grad_norm": 0.6806847453117371,
+ "learning_rate": 0.00014162162162162161,
+ "loss": 0.31959983706474304,
+ "num_input_tokens_seen": 59026,
+ "step": 62,
+ "train_runtime": 263.3802,
+ "train_tokens_per_second": 224.109
+ },
+ {
+ "epoch": 0.014911242603550296,
+ "grad_norm": 0.6916660070419312,
+ "learning_rate": 0.00014054054054054056,
+ "loss": 0.312305748462677,
+ "num_input_tokens_seen": 59982,
+ "step": 63,
+ "train_runtime": 266.3794,
+ "train_tokens_per_second": 225.175
+ },
+ {
+ "epoch": 0.01514792899408284,
+ "grad_norm": 0.5516266226768494,
+ "learning_rate": 0.00013945945945945947,
+ "loss": 0.2653953433036804,
+ "num_input_tokens_seen": 60926,
+ "step": 64,
+ "train_runtime": 269.3893,
+ "train_tokens_per_second": 226.163
+ },
+ {
+ "epoch": 0.015384615384615385,
+ "grad_norm": 0.5579270720481873,
+ "learning_rate": 0.0001383783783783784,
+ "loss": 0.20462818443775177,
+ "num_input_tokens_seen": 61844,
+ "step": 65,
+ "train_runtime": 272.3982,
+ "train_tokens_per_second": 227.035
+ },
+ {
+ "epoch": 0.015621301775147929,
+ "grad_norm": 0.7645660638809204,
+ "learning_rate": 0.0001372972972972973,
+ "loss": 0.2650893032550812,
+ "num_input_tokens_seen": 62766,
+ "step": 66,
+ "train_runtime": 275.404,
+ "train_tokens_per_second": 227.905
+ },
+ {
+ "epoch": 0.015857988165680473,
+ "grad_norm": 0.5974738001823425,
+ "learning_rate": 0.00013621621621621622,
+ "loss": 0.21247778832912445,
+ "num_input_tokens_seen": 63690,
+ "step": 67,
+ "train_runtime": 278.4021,
+ "train_tokens_per_second": 228.77
+ },
+ {
+ "epoch": 0.016094674556213016,
+ "grad_norm": 0.5831193327903748,
+ "learning_rate": 0.00013513513513513514,
+ "loss": 0.2024172842502594,
+ "num_input_tokens_seen": 64496,
+ "step": 68,
+ "train_runtime": 281.4313,
+ "train_tokens_per_second": 229.171
+ },
+ {
+ "epoch": 0.016331360946745564,
+ "grad_norm": 0.8678795695304871,
+ "learning_rate": 0.00013405405405405408,
+ "loss": 0.31490638852119446,
+ "num_input_tokens_seen": 65460,
+ "step": 69,
+ "train_runtime": 284.6137,
+ "train_tokens_per_second": 229.996
+ },
+ {
+ "epoch": 0.016568047337278107,
+ "grad_norm": 0.7611344456672668,
+ "learning_rate": 0.00013297297297297297,
+ "loss": 0.24661925435066223,
+ "num_input_tokens_seen": 66374,
+ "step": 70,
+ "train_runtime": 287.7742,
+ "train_tokens_per_second": 230.646
+ },
+ {
+ "epoch": 0.01680473372781065,
+ "grad_norm": 0.6685308814048767,
+ "learning_rate": 0.0001318918918918919,
+ "loss": 0.1843535602092743,
+ "num_input_tokens_seen": 67112,
+ "step": 71,
+ "train_runtime": 290.9321,
+ "train_tokens_per_second": 230.679
+ },
+ {
+ "epoch": 0.017041420118343194,
+ "grad_norm": 0.6150883436203003,
+ "learning_rate": 0.0001308108108108108,
+ "loss": 0.23372295498847961,
+ "num_input_tokens_seen": 68154,
+ "step": 72,
+ "train_runtime": 294.0809,
+ "train_tokens_per_second": 231.753
+ },
+ {
+ "epoch": 0.017278106508875738,
+ "grad_norm": 0.6938812732696533,
+ "learning_rate": 0.00012972972972972974,
+ "loss": 0.22619186341762543,
+ "num_input_tokens_seen": 69102,
+ "step": 73,
+ "train_runtime": 297.1843,
+ "train_tokens_per_second": 232.522
+ },
+ {
+ "epoch": 0.017514792899408285,
+ "grad_norm": 0.617787778377533,
+ "learning_rate": 0.00012864864864864866,
+ "loss": 0.2535328269004822,
+ "num_input_tokens_seen": 70010,
+ "step": 74,
+ "train_runtime": 300.2851,
+ "train_tokens_per_second": 233.145
+ },
+ {
+ "epoch": 0.01775147928994083,
+ "grad_norm": 0.6884900331497192,
+ "learning_rate": 0.00012756756756756758,
+ "loss": 0.21632739901542664,
+ "num_input_tokens_seen": 70940,
+ "step": 75,
+ "train_runtime": 303.3533,
+ "train_tokens_per_second": 233.853
+ },
+ {
+ "epoch": 0.017988165680473372,
+ "grad_norm": 0.5782189965248108,
+ "learning_rate": 0.0001264864864864865,
+ "loss": 0.20167675614356995,
+ "num_input_tokens_seen": 71862,
+ "step": 76,
+ "train_runtime": 306.4503,
+ "train_tokens_per_second": 234.498
+ },
+ {
+ "epoch": 0.018224852071005916,
+ "grad_norm": 0.6541483402252197,
+ "learning_rate": 0.0001254054054054054,
+ "loss": 0.23231221735477448,
+ "num_input_tokens_seen": 72812,
+ "step": 77,
+ "train_runtime": 309.5743,
+ "train_tokens_per_second": 235.2
+ },
+ {
+ "epoch": 0.018461538461538463,
+ "grad_norm": 0.7524886727333069,
+ "learning_rate": 0.00012432432432432433,
+ "loss": 0.27632662653923035,
+ "num_input_tokens_seen": 73774,
+ "step": 78,
+ "train_runtime": 312.6963,
+ "train_tokens_per_second": 235.929
+ },
+ {
+ "epoch": 0.018698224852071007,
+ "grad_norm": 0.716282308101654,
+ "learning_rate": 0.00012324324324324327,
+ "loss": 0.2734939754009247,
+ "num_input_tokens_seen": 74716,
+ "step": 79,
+ "train_runtime": 315.7612,
+ "train_tokens_per_second": 236.622
+ },
+ {
+ "epoch": 0.01893491124260355,
+ "grad_norm": 0.7010631561279297,
+ "learning_rate": 0.00012216216216216216,
+ "loss": 0.2774362564086914,
+ "num_input_tokens_seen": 75692,
+ "step": 80,
+ "train_runtime": 318.7983,
+ "train_tokens_per_second": 237.429
+ },
+ {
+ "epoch": 0.019171597633136094,
+ "grad_norm": 0.6561240553855896,
+ "learning_rate": 0.00012108108108108109,
+ "loss": 0.21640686690807343,
+ "num_input_tokens_seen": 76558,
+ "step": 81,
+ "train_runtime": 321.889,
+ "train_tokens_per_second": 237.84
+ },
+ {
+ "epoch": 0.019408284023668638,
+ "grad_norm": 0.5552831888198853,
+ "learning_rate": 0.00012,
+ "loss": 0.21232956647872925,
+ "num_input_tokens_seen": 77408,
+ "step": 82,
+ "train_runtime": 325.0469,
+ "train_tokens_per_second": 238.144
+ },
+ {
+ "epoch": 0.019644970414201185,
+ "grad_norm": 0.5905120372772217,
+ "learning_rate": 0.00011891891891891893,
+ "loss": 0.22443550825119019,
+ "num_input_tokens_seen": 78278,
+ "step": 83,
+ "train_runtime": 328.099,
+ "train_tokens_per_second": 238.58
+ },
+ {
+ "epoch": 0.01988165680473373,
+ "grad_norm": 0.6606366634368896,
+ "learning_rate": 0.00011783783783783784,
+ "loss": 0.2400747388601303,
+ "num_input_tokens_seen": 79262,
+ "step": 84,
+ "train_runtime": 331.1361,
+ "train_tokens_per_second": 239.364
+ },
+ {
+ "epoch": 0.020118343195266272,
+ "grad_norm": 0.6508055329322815,
+ "learning_rate": 0.00011675675675675676,
+ "loss": 0.19072499871253967,
+ "num_input_tokens_seen": 80144,
+ "step": 85,
+ "train_runtime": 334.2158,
+ "train_tokens_per_second": 239.797
+ },
+ {
+ "epoch": 0.020355029585798816,
+ "grad_norm": 0.606562614440918,
+ "learning_rate": 0.00011567567567567568,
+ "loss": 0.20951862633228302,
+ "num_input_tokens_seen": 81108,
+ "step": 86,
+ "train_runtime": 337.3105,
+ "train_tokens_per_second": 240.455
+ },
+ {
+ "epoch": 0.02059171597633136,
+ "grad_norm": 0.6757521629333496,
+ "learning_rate": 0.00011459459459459461,
+ "loss": 0.17614495754241943,
+ "num_input_tokens_seen": 82032,
+ "step": 87,
+ "train_runtime": 340.3438,
+ "train_tokens_per_second": 241.027
+ },
+ {
+ "epoch": 0.020828402366863907,
+ "grad_norm": 0.671481728553772,
+ "learning_rate": 0.00011351351351351351,
+ "loss": 0.1901710331439972,
+ "num_input_tokens_seen": 82954,
+ "step": 88,
+ "train_runtime": 343.4059,
+ "train_tokens_per_second": 241.563
+ },
+ {
+ "epoch": 0.02106508875739645,
+ "grad_norm": 0.5919027924537659,
+ "learning_rate": 0.00011243243243243244,
+ "loss": 0.1563737988471985,
+ "num_input_tokens_seen": 83880,
+ "step": 89,
+ "train_runtime": 346.5125,
+ "train_tokens_per_second": 242.069
+ },
+ {
+ "epoch": 0.021301775147928994,
+ "grad_norm": 0.7108309268951416,
+ "learning_rate": 0.00011135135135135135,
+ "loss": 0.17256894707679749,
+ "num_input_tokens_seen": 84852,
+ "step": 90,
+ "train_runtime": 349.6508,
+ "train_tokens_per_second": 242.676
+ },
+ {
+ "epoch": 0.021538461538461538,
+ "grad_norm": 0.8608420491218567,
+ "learning_rate": 0.00011027027027027029,
+ "loss": 0.1845530867576599,
+ "num_input_tokens_seen": 85820,
+ "step": 91,
+ "train_runtime": 356.267,
+ "train_tokens_per_second": 240.887
+ },
+ {
+ "epoch": 0.02177514792899408,
+ "grad_norm": 0.9677010774612427,
+ "learning_rate": 0.00010918918918918919,
+ "loss": 0.24358105659484863,
+ "num_input_tokens_seen": 86770,
+ "step": 92,
+ "train_runtime": 359.3722,
+ "train_tokens_per_second": 241.449
+ },
+ {
+ "epoch": 0.02201183431952663,
+ "grad_norm": 0.6302966475486755,
+ "learning_rate": 0.00010810810810810812,
+ "loss": 0.18155133724212646,
+ "num_input_tokens_seen": 87736,
+ "step": 93,
+ "train_runtime": 362.5874,
+ "train_tokens_per_second": 241.972
+ },
+ {
+ "epoch": 0.022248520710059172,
+ "grad_norm": 0.7171812653541565,
+ "learning_rate": 0.00010702702702702702,
+ "loss": 0.15349116921424866,
+ "num_input_tokens_seen": 88626,
+ "step": 94,
+ "train_runtime": 365.7055,
+ "train_tokens_per_second": 242.343
+ },
+ {
+ "epoch": 0.022485207100591716,
+ "grad_norm": 0.7326887845993042,
+ "learning_rate": 0.00010594594594594595,
+ "loss": 0.18243758380413055,
+ "num_input_tokens_seen": 89548,
+ "step": 95,
+ "train_runtime": 368.799,
+ "train_tokens_per_second": 242.81
+ },
+ {
+ "epoch": 0.02272189349112426,
+ "grad_norm": 0.6686019897460938,
+ "learning_rate": 0.00010486486486486487,
+ "loss": 0.15622465312480927,
+ "num_input_tokens_seen": 90504,
+ "step": 96,
+ "train_runtime": 371.8404,
+ "train_tokens_per_second": 243.395
+ },
+ {
+ "epoch": 0.022958579881656806,
+ "grad_norm": 0.6306503415107727,
+ "learning_rate": 0.0001037837837837838,
+ "loss": 0.1272117793560028,
+ "num_input_tokens_seen": 91454,
+ "step": 97,
+ "train_runtime": 375.0199,
+ "train_tokens_per_second": 243.864
+ },
+ {
+ "epoch": 0.02319526627218935,
+ "grad_norm": 0.6068493127822876,
+ "learning_rate": 0.0001027027027027027,
+ "loss": 0.1577155441045761,
+ "num_input_tokens_seen": 92466,
+ "step": 98,
+ "train_runtime": 378.0978,
+ "train_tokens_per_second": 244.556
+ },
+ {
+ "epoch": 0.023431952662721894,
+ "grad_norm": 0.7263635993003845,
+ "learning_rate": 0.00010162162162162163,
+ "loss": 0.1845340132713318,
+ "num_input_tokens_seen": 93398,
+ "step": 99,
+ "train_runtime": 381.1358,
+ "train_tokens_per_second": 245.052
+ },
+ {
+ "epoch": 0.023668639053254437,
+ "grad_norm": 0.6466794013977051,
+ "learning_rate": 0.00010054054054054053,
+ "loss": 0.14443959295749664,
+ "num_input_tokens_seen": 94230,
+ "step": 100,
+ "train_runtime": 384.2906,
+ "train_tokens_per_second": 245.205
+ },
+ {
+ "epoch": 0.02390532544378698,
+ "grad_norm": 0.6924043297767639,
+ "learning_rate": 9.945945945945948e-05,
+ "loss": 0.18473166227340698,
+ "num_input_tokens_seen": 95132,
+ "step": 101,
+ "train_runtime": 387.5005,
+ "train_tokens_per_second": 245.502
+ },
+ {
+ "epoch": 0.024142011834319528,
+ "grad_norm": 0.732337236404419,
+ "learning_rate": 9.837837837837839e-05,
+ "loss": 0.16353662312030792,
+ "num_input_tokens_seen": 96080,
+ "step": 102,
+ "train_runtime": 390.584,
+ "train_tokens_per_second": 245.991
+ },
+ {
+ "epoch": 0.02437869822485207,
+ "grad_norm": 0.5956932902336121,
+ "learning_rate": 9.729729729729731e-05,
+ "loss": 0.16359847784042358,
+ "num_input_tokens_seen": 97048,
+ "step": 103,
+ "train_runtime": 393.6474,
+ "train_tokens_per_second": 246.535
+ },
+ {
+ "epoch": 0.024615384615384615,
+ "grad_norm": 0.6137388348579407,
+ "learning_rate": 9.621621621621622e-05,
+ "loss": 0.15779417753219604,
+ "num_input_tokens_seen": 97986,
+ "step": 104,
+ "train_runtime": 396.7042,
+ "train_tokens_per_second": 247.0
+ },
+ {
+ "epoch": 0.02485207100591716,
+ "grad_norm": 0.7936599254608154,
+ "learning_rate": 9.513513513513514e-05,
+ "loss": 0.21357379853725433,
+ "num_input_tokens_seen": 98894,
+ "step": 105,
+ "train_runtime": 399.8228,
+ "train_tokens_per_second": 247.345
+ },
+ {
+ "epoch": 0.025088757396449703,
+ "grad_norm": 0.6574507355690002,
+ "learning_rate": 9.405405405405407e-05,
+ "loss": 0.1685573309659958,
+ "num_input_tokens_seen": 99856,
+ "step": 106,
+ "train_runtime": 402.8867,
+ "train_tokens_per_second": 247.851
+ },
+ {
+ "epoch": 0.02532544378698225,
+ "grad_norm": 0.5891634821891785,
+ "learning_rate": 9.297297297297299e-05,
+ "loss": 0.12082140147686005,
+ "num_input_tokens_seen": 100810,
+ "step": 107,
+ "train_runtime": 405.9752,
+ "train_tokens_per_second": 248.316
+ },
+ {
+ "epoch": 0.025562130177514793,
+ "grad_norm": 0.6624342799186707,
+ "learning_rate": 9.18918918918919e-05,
+ "loss": 0.1473795771598816,
+ "num_input_tokens_seen": 101744,
+ "step": 108,
+ "train_runtime": 409.0084,
+ "train_tokens_per_second": 248.758
+ },
+ {
+ "epoch": 0.025798816568047337,
+ "grad_norm": 0.5968036651611328,
+ "learning_rate": 9.081081081081082e-05,
+ "loss": 0.12923751771450043,
+ "num_input_tokens_seen": 102702,
+ "step": 109,
+ "train_runtime": 412.199,
+ "train_tokens_per_second": 249.156
+ },
+ {
+ "epoch": 0.02603550295857988,
+ "grad_norm": 0.5954580307006836,
+ "learning_rate": 8.972972972972973e-05,
+ "loss": 0.1161399856209755,
+ "num_input_tokens_seen": 103672,
+ "step": 110,
+ "train_runtime": 415.3736,
+ "train_tokens_per_second": 249.587
+ },
+ {
+ "epoch": 0.026272189349112424,
+ "grad_norm": 0.6139379739761353,
+ "learning_rate": 8.864864864864866e-05,
+ "loss": 0.12352899461984634,
+ "num_input_tokens_seen": 104598,
+ "step": 111,
+ "train_runtime": 418.3271,
+ "train_tokens_per_second": 250.039
+ },
+ {
+ "epoch": 0.02650887573964497,
+ "grad_norm": 0.7933095097541809,
+ "learning_rate": 8.756756756756758e-05,
+ "loss": 0.1709994524717331,
+ "num_input_tokens_seen": 105580,
+ "step": 112,
+ "train_runtime": 421.3565,
+ "train_tokens_per_second": 250.572
+ },
+ {
+ "epoch": 0.026745562130177515,
+ "grad_norm": 0.7641370892524719,
+ "learning_rate": 8.64864864864865e-05,
+ "loss": 0.13086311519145966,
+ "num_input_tokens_seen": 106444,
+ "step": 113,
+ "train_runtime": 424.4668,
+ "train_tokens_per_second": 250.771
+ },
+ {
+ "epoch": 0.02698224852071006,
+ "grad_norm": 0.7611938118934631,
+ "learning_rate": 8.540540540540541e-05,
+ "loss": 0.16537640988826752,
+ "num_input_tokens_seen": 107454,
+ "step": 114,
+ "train_runtime": 427.5,
+ "train_tokens_per_second": 251.354
+ },
+ {
+ "epoch": 0.027218934911242602,
+ "grad_norm": 0.7495304346084595,
+ "learning_rate": 8.432432432432433e-05,
+ "loss": 0.14228317141532898,
+ "num_input_tokens_seen": 108426,
+ "step": 115,
+ "train_runtime": 430.5035,
+ "train_tokens_per_second": 251.859
+ },
+ {
+ "epoch": 0.02745562130177515,
+ "grad_norm": 1.2078282833099365,
+ "learning_rate": 8.324324324324326e-05,
+ "loss": 0.10530930012464523,
+ "num_input_tokens_seen": 109446,
+ "step": 116,
+ "train_runtime": 433.4964,
+ "train_tokens_per_second": 252.473
+ },
+ {
+ "epoch": 0.027692307692307693,
+ "grad_norm": 0.6765572428703308,
+ "learning_rate": 8.216216216216217e-05,
+ "loss": 0.10717278718948364,
+ "num_input_tokens_seen": 110374,
+ "step": 117,
+ "train_runtime": 436.5901,
+ "train_tokens_per_second": 252.809
+ },
+ {
+ "epoch": 0.027928994082840237,
+ "grad_norm": 0.8501551151275635,
+ "learning_rate": 8.108108108108109e-05,
+ "loss": 0.17784273624420166,
+ "num_input_tokens_seen": 111340,
+ "step": 118,
+ "train_runtime": 439.6526,
+ "train_tokens_per_second": 253.245
+ },
+ {
+ "epoch": 0.02816568047337278,
+ "grad_norm": 0.7583926320075989,
+ "learning_rate": 8e-05,
+ "loss": 0.16588062047958374,
+ "num_input_tokens_seen": 112288,
+ "step": 119,
+ "train_runtime": 442.8046,
+ "train_tokens_per_second": 253.584
+ },
+ {
+ "epoch": 0.028402366863905324,
+ "grad_norm": 0.6620173454284668,
+ "learning_rate": 7.891891891891892e-05,
+ "loss": 0.11676295101642609,
+ "num_input_tokens_seen": 113206,
+ "step": 120,
+ "train_runtime": 445.9682,
+ "train_tokens_per_second": 253.843
+ },
+ {
+ "epoch": 0.02863905325443787,
+ "grad_norm": 0.6844660639762878,
+ "learning_rate": 7.783783783783785e-05,
+ "loss": 0.09329869598150253,
+ "num_input_tokens_seen": 114134,
+ "step": 121,
+ "train_runtime": 452.5692,
+ "train_tokens_per_second": 252.191
+ },
+ {
+ "epoch": 0.028875739644970415,
+ "grad_norm": 0.7968376874923706,
+ "learning_rate": 7.675675675675677e-05,
+ "loss": 0.13320565223693848,
+ "num_input_tokens_seen": 115088,
+ "step": 122,
+ "train_runtime": 455.6245,
+ "train_tokens_per_second": 252.594
+ },
+ {
+ "epoch": 0.02911242603550296,
+ "grad_norm": 0.8079301118850708,
+ "learning_rate": 7.567567567567568e-05,
+ "loss": 0.1387016326189041,
+ "num_input_tokens_seen": 116006,
+ "step": 123,
+ "train_runtime": 458.7601,
+ "train_tokens_per_second": 252.869
+ },
+ {
+ "epoch": 0.029349112426035502,
+ "grad_norm": 1.1301748752593994,
+ "learning_rate": 7.45945945945946e-05,
+ "loss": 0.197790265083313,
+ "num_input_tokens_seen": 116958,
+ "step": 124,
+ "train_runtime": 461.861,
+ "train_tokens_per_second": 253.232
+ },
+ {
+ "epoch": 0.029585798816568046,
+ "grad_norm": 0.7272042632102966,
+ "learning_rate": 7.351351351351352e-05,
+ "loss": 0.12121891975402832,
+ "num_input_tokens_seen": 117886,
+ "step": 125,
+ "train_runtime": 465.0328,
+ "train_tokens_per_second": 253.5
+ },
+ {
+ "epoch": 0.029822485207100593,
+ "grad_norm": 0.6021063923835754,
+ "learning_rate": 7.243243243243245e-05,
+ "loss": 0.08234579861164093,
+ "num_input_tokens_seen": 118806,
+ "step": 126,
+ "train_runtime": 468.1518,
+ "train_tokens_per_second": 253.777
+ },
+ {
+ "epoch": 0.030059171597633137,
+ "grad_norm": 0.5659137964248657,
+ "learning_rate": 7.135135135135136e-05,
+ "loss": 0.07469362020492554,
+ "num_input_tokens_seen": 119716,
+ "step": 127,
+ "train_runtime": 471.2559,
+ "train_tokens_per_second": 254.036
+ },
+ {
+ "epoch": 0.03029585798816568,
+ "grad_norm": 0.7604466080665588,
+ "learning_rate": 7.027027027027028e-05,
+ "loss": 0.15167035162448883,
+ "num_input_tokens_seen": 120710,
+ "step": 128,
+ "train_runtime": 474.3536,
+ "train_tokens_per_second": 254.473
+ },
+ {
+ "epoch": 0.030532544378698224,
+ "grad_norm": 0.898281455039978,
+ "learning_rate": 6.91891891891892e-05,
+ "loss": 0.18358615040779114,
+ "num_input_tokens_seen": 121668,
+ "step": 129,
+ "train_runtime": 477.4676,
+ "train_tokens_per_second": 254.819
+ },
+ {
+ "epoch": 0.03076923076923077,
+ "grad_norm": 1.046539545059204,
+ "learning_rate": 6.810810810810811e-05,
+ "loss": 0.14173445105552673,
+ "num_input_tokens_seen": 122654,
+ "step": 130,
+ "train_runtime": 480.6242,
+ "train_tokens_per_second": 255.197
+ },
+ {
+ "epoch": 0.031005917159763315,
+ "grad_norm": 0.689860463142395,
+ "learning_rate": 6.702702702702704e-05,
+ "loss": 0.1232977956533432,
+ "num_input_tokens_seen": 123610,
+ "step": 131,
+ "train_runtime": 483.7904,
+ "train_tokens_per_second": 255.503
+ },
+ {
+ "epoch": 0.031242603550295858,
+ "grad_norm": 0.6417210698127747,
+ "learning_rate": 6.594594594594596e-05,
+ "loss": 0.08111131936311722,
+ "num_input_tokens_seen": 124526,
+ "step": 132,
+ "train_runtime": 486.9241,
+ "train_tokens_per_second": 255.74
+ },
+ {
+ "epoch": 0.0314792899408284,
+ "grad_norm": 0.8019670844078064,
+ "learning_rate": 6.486486486486487e-05,
+ "loss": 0.13235916197299957,
+ "num_input_tokens_seen": 125438,
+ "step": 133,
+ "train_runtime": 490.0416,
+ "train_tokens_per_second": 255.974
+ },
+ {
+ "epoch": 0.031715976331360946,
+ "grad_norm": 0.8618372082710266,
+ "learning_rate": 6.378378378378379e-05,
+ "loss": 0.11209192872047424,
+ "num_input_tokens_seen": 126378,
+ "step": 134,
+ "train_runtime": 493.0816,
+ "train_tokens_per_second": 256.302
+ },
+ {
+ "epoch": 0.03195266272189349,
+ "grad_norm": 0.6566123962402344,
+ "learning_rate": 6.27027027027027e-05,
+ "loss": 0.08107412606477737,
+ "num_input_tokens_seen": 127280,
+ "step": 135,
+ "train_runtime": 496.1373,
+ "train_tokens_per_second": 256.542
+ },
+ {
+ "epoch": 0.03218934911242603,
+ "grad_norm": 0.671886682510376,
+ "learning_rate": 6.162162162162163e-05,
+ "loss": 0.10395485162734985,
+ "num_input_tokens_seen": 128246,
+ "step": 136,
+ "train_runtime": 499.2198,
+ "train_tokens_per_second": 256.893
+ },
+ {
+ "epoch": 0.032426035502958576,
+ "grad_norm": 0.5853029489517212,
+ "learning_rate": 6.0540540540540543e-05,
+ "loss": 0.06830425560474396,
+ "num_input_tokens_seen": 129142,
+ "step": 137,
+ "train_runtime": 502.2719,
+ "train_tokens_per_second": 257.116
+ },
+ {
+ "epoch": 0.03266272189349113,
+ "grad_norm": 0.9871621131896973,
+ "learning_rate": 5.9459459459459466e-05,
+ "loss": 0.14138084650039673,
+ "num_input_tokens_seen": 130060,
+ "step": 138,
+ "train_runtime": 505.3687,
+ "train_tokens_per_second": 257.357
+ },
+ {
+ "epoch": 0.03289940828402367,
+ "grad_norm": 0.6716732978820801,
+ "learning_rate": 5.837837837837838e-05,
+ "loss": 0.07525799423456192,
+ "num_input_tokens_seen": 130982,
+ "step": 139,
+ "train_runtime": 508.4248,
+ "train_tokens_per_second": 257.623
+ },
+ {
+ "epoch": 0.033136094674556214,
+ "grad_norm": 0.7177789807319641,
+ "learning_rate": 5.7297297297297305e-05,
+ "loss": 0.09517940878868103,
+ "num_input_tokens_seen": 131900,
+ "step": 140,
+ "train_runtime": 511.57,
+ "train_tokens_per_second": 257.834
+ },
+ {
+ "epoch": 0.03337278106508876,
+ "grad_norm": 0.703551709651947,
+ "learning_rate": 5.621621621621622e-05,
+ "loss": 0.09361469000577927,
+ "num_input_tokens_seen": 132890,
+ "step": 141,
+ "train_runtime": 514.7186,
+ "train_tokens_per_second": 258.18
+ },
+ {
+ "epoch": 0.0336094674556213,
+ "grad_norm": 0.7156357765197754,
+ "learning_rate": 5.5135135135135144e-05,
+ "loss": 0.08832181245088577,
+ "num_input_tokens_seen": 133810,
+ "step": 142,
+ "train_runtime": 517.7644,
+ "train_tokens_per_second": 258.438
+ },
+ {
+ "epoch": 0.033846153846153845,
+ "grad_norm": 0.7757886648178101,
+ "learning_rate": 5.405405405405406e-05,
+ "loss": 0.12478049844503403,
+ "num_input_tokens_seen": 134758,
+ "step": 143,
+ "train_runtime": 520.8215,
+ "train_tokens_per_second": 258.741
+ },
+ {
+ "epoch": 0.03408284023668639,
+ "grad_norm": 0.6047410368919373,
+ "learning_rate": 5.2972972972972976e-05,
+ "loss": 0.07645335048437119,
+ "num_input_tokens_seen": 135724,
+ "step": 144,
+ "train_runtime": 523.9787,
+ "train_tokens_per_second": 259.026
+ },
+ {
+ "epoch": 0.03431952662721893,
+ "grad_norm": 0.7750188708305359,
+ "learning_rate": 5.18918918918919e-05,
+ "loss": 0.10904290527105331,
+ "num_input_tokens_seen": 136662,
+ "step": 145,
+ "train_runtime": 527.0633,
+ "train_tokens_per_second": 259.29
+ },
+ {
+ "epoch": 0.034556213017751476,
+ "grad_norm": 0.6400149464607239,
+ "learning_rate": 5.0810810810810815e-05,
+ "loss": 0.07525563985109329,
+ "num_input_tokens_seen": 137628,
+ "step": 146,
+ "train_runtime": 530.1299,
+ "train_tokens_per_second": 259.612
+ },
+ {
+ "epoch": 0.03479289940828403,
+ "grad_norm": 0.6681646108627319,
+ "learning_rate": 4.972972972972974e-05,
+ "loss": 0.08536849915981293,
+ "num_input_tokens_seen": 138610,
+ "step": 147,
+ "train_runtime": 533.159,
+ "train_tokens_per_second": 259.979
+ },
+ {
+ "epoch": 0.03502958579881657,
+ "grad_norm": 0.6158885359764099,
+ "learning_rate": 4.8648648648648654e-05,
+ "loss": 0.0753958523273468,
+ "num_input_tokens_seen": 139486,
+ "step": 148,
+ "train_runtime": 536.2632,
+ "train_tokens_per_second": 260.107
+ },
+ {
+ "epoch": 0.035266272189349114,
+ "grad_norm": 0.665565550327301,
+ "learning_rate": 4.756756756756757e-05,
+ "loss": 0.08198000490665436,
+ "num_input_tokens_seen": 140440,
+ "step": 149,
+ "train_runtime": 539.3165,
+ "train_tokens_per_second": 260.404
+ },
+ {
+ "epoch": 0.03550295857988166,
+ "grad_norm": 0.6818353533744812,
+ "learning_rate": 4.648648648648649e-05,
+ "loss": 0.0710655227303505,
+ "num_input_tokens_seen": 141344,
+ "step": 150,
+ "train_runtime": 542.3824,
+ "train_tokens_per_second": 260.598
+ }
+ ],
+ "logging_steps": 1,
+ "max_steps": 190,
+ "num_input_tokens_seen": 141344,
+ "num_train_epochs": 1,
+ "save_steps": 30,
+ "stateful_callbacks": {
+ "TrainerControl": {
+ "args": {
+ "should_epoch_stop": false,
+ "should_evaluate": false,
+ "should_log": false,
+ "should_save": true,
+ "should_training_stop": false
+ },
+ "attributes": {}
+ }
+ },
+ "total_flos": 2037689236101120.0,
+ "train_batch_size": 2,
+ "trial_name": null,
+ "trial_params": null
+}
diff --git a/checkpoint-150/training_args.bin b/checkpoint-150/training_args.bin
new file mode 100644
index 0000000000000000000000000000000000000000..cefffe2bb567bdf18003e5cc5aaf746472c2981f
--- /dev/null
+++ b/checkpoint-150/training_args.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:34c4ca6d50f04301fb38588516a91421e14a8ad1ffddf12afdfd9ba35cd4b64e
+size 5777
diff --git a/checkpoint-180/README.md b/checkpoint-180/README.md
new file mode 100644
index 0000000000000000000000000000000000000000..41470ec6557bbc37155893f70bd59986401b9148
--- /dev/null
+++ b/checkpoint-180/README.md
@@ -0,0 +1,210 @@
+---
+base_model: unsloth/gemma-4-e2b-it-unsloth-bnb-4bit
+library_name: peft
+pipeline_tag: text-generation
+tags:
+- base_model:adapter:unsloth/gemma-4-e2b-it-unsloth-bnb-4bit
+- lora
+- sft
+- transformers
+- trl
+- unsloth
+---
+
+# Model Card for Model ID
+
+
+
+
+
+## Model Details
+
+### Model Description
+
+
+
+
+
+- **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]
+
+
+
+- **Repository:** [More Information Needed]
+- **Paper [optional]:** [More Information Needed]
+- **Demo [optional]:** [More Information Needed]
+
+## Uses
+
+
+
+### Direct Use
+
+
+
+[More Information Needed]
+
+### Downstream Use [optional]
+
+
+
+[More Information Needed]
+
+### Out-of-Scope Use
+
+
+
+[More Information Needed]
+
+## Bias, Risks, and Limitations
+
+
+
+[More Information Needed]
+
+### Recommendations
+
+
+
+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
+
+
+
+[More Information Needed]
+
+### Training Procedure
+
+
+
+#### Preprocessing [optional]
+
+[More Information Needed]
+
+
+#### Training Hyperparameters
+
+- **Training regime:** [More Information Needed]
+
+#### Speeds, Sizes, Times [optional]
+
+
+
+[More Information Needed]
+
+## Evaluation
+
+
+
+### Testing Data, Factors & Metrics
+
+#### Testing Data
+
+
+
+[More Information Needed]
+
+#### Factors
+
+
+
+[More Information Needed]
+
+#### Metrics
+
+
+
+[More Information Needed]
+
+### Results
+
+[More Information Needed]
+
+#### Summary
+
+
+
+## Model Examination [optional]
+
+
+
+[More Information Needed]
+
+## Environmental Impact
+
+
+
+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]
+
+
+
+**BibTeX:**
+
+[More Information Needed]
+
+**APA:**
+
+[More Information Needed]
+
+## Glossary [optional]
+
+
+
+[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
\ No newline at end of file
diff --git a/checkpoint-180/adapter_config.json b/checkpoint-180/adapter_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..915726be1cd67eb165724a8390b51a57fe0dd6df
--- /dev/null
+++ b/checkpoint-180/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": "unsloth/gemma-4-e2b-it-unsloth-bnb-4bit",
+ "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": [
+ "q_proj",
+ "gate_proj",
+ "v_proj",
+ "k_proj",
+ "down_proj",
+ "o_proj",
+ "up_proj"
+ ],
+ "target_parameters": null,
+ "task_type": "CAUSAL_LM",
+ "trainable_token_indices": null,
+ "use_dora": false,
+ "use_qalora": false,
+ "use_rslora": false
+}
\ No newline at end of file
diff --git a/checkpoint-180/adapter_model.safetensors b/checkpoint-180/adapter_model.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..0955267837a52e4d43cba77049e357c2e2051f25
--- /dev/null
+++ b/checkpoint-180/adapter_model.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4e2c1029454f1834f511c4c2d3a52d76531655516a9d3045514d762b5156feb6
+size 124277728
diff --git a/checkpoint-180/chat_template.jinja b/checkpoint-180/chat_template.jinja
new file mode 100644
index 0000000000000000000000000000000000000000..dc032e46e28e0d325b18d26da1279cdda3afa8d7
--- /dev/null
+++ b/checkpoint-180/chat_template.jinja
@@ -0,0 +1,351 @@
+{%- macro format_parameters(properties, required, filter_keys=false) -%}
+ {%- 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 not filter_keys or 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([]), filter_keys=true) -}}
+ }
+ {%- 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('') -%}
+ {%- 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 -%}
+ {{- '' -}}
+{%- 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'] -%}
+ {%- if messages[0]['content'] is string -%}
+ {{- messages[0]['content'] | trim -}}
+ {%- elif messages[0]['content'] is sequence -%}
+ {%- for item in messages[0]['content'] -%}
+ {{- item['text'] | trim + ' '-}}
+ {%- endfor -%}
+ {%- endif -%}
+ {%- set loop_messages = messages[1:] -%}
+ {%- endif -%}
+ {%- if tools -%}
+ {%- for tool in tools %}
+ {{- '<|tool>' -}}
+ {{- format_function_declaration(tool) | trim -}}
+ {{- '' -}}
+ {%- endfor %}
+ {%- set ns.prev_message_type = 'tool' -%}
+ {%- endif -%}
+ {{- '\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' -}}
+ {%- 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 -%}
+ {{- '}' -}}
+ {%- 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 -%}
+
+ {%- set captured_content -%}
+ {%- 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 -%}
+ {%- endset -%}
+
+ {{- captured_content -}}
+ {%- set has_content = captured_content | trim | length > 0 -%}
+
+ {%- if ns.prev_message_type == 'tool_call' and not ns_tr_out.flag -%}
+ {{- '<|tool_response>' -}}
+ {%- elif not (ns_tr_out.flag and not has_content) -%}
+ {{- '\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 -%}
\ No newline at end of file
diff --git a/checkpoint-180/optimizer.pt b/checkpoint-180/optimizer.pt
new file mode 100644
index 0000000000000000000000000000000000000000..396b6690f525bbe940e019f43b8da5834f908301
--- /dev/null
+++ b/checkpoint-180/optimizer.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d28a0dc846feefdc15b5a30c4ff928dcdd8416bb2c13beef4fa5f4ca242bd821
+size 52047117
diff --git a/checkpoint-180/processor_config.json b/checkpoint-180/processor_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..bfaa5e360ad17c2761b7ee3b05abf03e381e8aa1
--- /dev/null
+++ b/checkpoint-180/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"
+ }
+}
diff --git a/checkpoint-180/rng_state.pth b/checkpoint-180/rng_state.pth
new file mode 100644
index 0000000000000000000000000000000000000000..3ef66339b9befa098183fd5d69faed6838e526b0
--- /dev/null
+++ b/checkpoint-180/rng_state.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f1d565802a8e26c4e8a31328752b7a7fdc186d9401aa008e65697d0ad8c22e33
+size 14645
diff --git a/checkpoint-180/scaler.pt b/checkpoint-180/scaler.pt
new file mode 100644
index 0000000000000000000000000000000000000000..6d1a2b946581ac44713999963c836d5ea2df6738
--- /dev/null
+++ b/checkpoint-180/scaler.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:157cc77e7dd6e9c2bce809a6b120c6be65dd2e62a42b62dfda1d462877a40ec3
+size 1383
diff --git a/checkpoint-180/scheduler.pt b/checkpoint-180/scheduler.pt
new file mode 100644
index 0000000000000000000000000000000000000000..5db1001aabf925209d186bbb6f5b0edef90b6ef7
--- /dev/null
+++ b/checkpoint-180/scheduler.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ec2bd51ad4fa32551317ffe221c08d896ed511f96c24b01b67eb1975b12513fc
+size 1465
diff --git a/checkpoint-180/tokenizer.json b/checkpoint-180/tokenizer.json
new file mode 100644
index 0000000000000000000000000000000000000000..1ff9f3e3439a939b971f9919e821bf87e835a503
--- /dev/null
+++ b/checkpoint-180/tokenizer.json
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:cc8d3a0ce36466ccc1278bf987df5f71db1719b9ca6b4118264f45cb627bfe0f
+size 32169626
diff --git a/checkpoint-180/tokenizer_config.json b/checkpoint-180/tokenizer_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..24b1522bb32b0eb05b631b1a9f112751b8cb0fab
--- /dev/null
+++ b/checkpoint-180/tokenizer_config.json
@@ -0,0 +1,289 @@
+{
+ "audio_token": "<|audio|>",
+ "backend": "tokenizers",
+ "boa_token": "<|audio>",
+ "boi_token": "<|image>",
+ "bos_token": "",
+ "eoa_token": "",
+ "eoc_token": "",
+ "eoi_token": "",
+ "eos_token": "",
+ "eot_token": "",
+ "escape_token": "<|\"|>",
+ "etc_token": "",
+ "etd_token": "",
+ "etr_token": "",
+ "extra_special_tokens": [
+ "<|video|>"
+ ],
+ "image_token": "<|image|>",
+ "is_local": false,
+ "mask_token": "",
+ "model_max_length": 131072,
+ "model_specific_special_tokens": {
+ "audio_token": "<|audio|>",
+ "boa_token": "<|audio>",
+ "boi_token": "<|image>",
+ "eoa_token": "",
+ "eoc_token": "",
+ "eoi_token": "",
+ "eot_token": "",
+ "escape_token": "<|\"|>",
+ "etc_token": "",
+ "etd_token": "",
+ "etr_token": "",
+ "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": "",
+ "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\\w+)(?P\\{.*\\})"
+ },
+ "type": {
+ "const": "function"
+ }
+ },
+ "type": "object"
+ },
+ "type": "array",
+ "x-regex-iterator": "<\\|tool_call>(.*?)"
+ }
+ },
+ "type": "object",
+ "x-regex": "(\\<\\|channel\\>thought\\n(?P.*?)\\)?(?P\\<\\|tool_call\\>.*\\)?(?P(?:(?!\\)(?!\\<\\|tool_response\\>).)+)?(?:\\|\\<\\|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": "",
+ "added_tokens_decoder": {
+ "0": {
+ "content": "",
+ "single_word": false,
+ "lstrip": false,
+ "rstrip": false,
+ "normalized": false,
+ "special": true
+ },
+ "1": {
+ "content": "",
+ "single_word": false,
+ "lstrip": false,
+ "rstrip": false,
+ "normalized": false,
+ "special": true
+ },
+ "2": {
+ "content": "",
+ "single_word": false,
+ "lstrip": false,
+ "rstrip": false,
+ "normalized": false,
+ "special": true
+ },
+ "3": {
+ "content": "",
+ "single_word": false,
+ "lstrip": false,
+ "rstrip": false,
+ "normalized": false,
+ "special": true
+ },
+ "4": {
+ "content": "",
+ "single_word": false,
+ "lstrip": false,
+ "rstrip": false,
+ "normalized": false,
+ "special": true
+ },
+ "46": {
+ "content": "<|tool>",
+ "single_word": false,
+ "lstrip": false,
+ "rstrip": false,
+ "normalized": false,
+ "special": true
+ },
+ "47": {
+ "content": "",
+ "single_word": false,
+ "lstrip": false,
+ "rstrip": false,
+ "normalized": false,
+ "special": true
+ },
+ "48": {
+ "content": "<|tool_call>",
+ "single_word": false,
+ "lstrip": false,
+ "rstrip": false,
+ "normalized": false,
+ "special": true
+ },
+ "49": {
+ "content": "",
+ "single_word": false,
+ "lstrip": false,
+ "rstrip": false,
+ "normalized": false,
+ "special": true
+ },
+ "50": {
+ "content": "<|tool_response>",
+ "single_word": false,
+ "lstrip": false,
+ "rstrip": false,
+ "normalized": false,
+ "special": true
+ },
+ "51": {
+ "content": "",
+ "single_word": false,
+ "lstrip": false,
+ "rstrip": false,
+ "normalized": false,
+ "special": true
+ },
+ "52": {
+ "content": "<|\"|>",
+ "single_word": false,
+ "lstrip": false,
+ "rstrip": false,
+ "normalized": false,
+ "special": true
+ },
+ "98": {
+ "content": "<|think|>",
+ "single_word": false,
+ "lstrip": false,
+ "rstrip": false,
+ "normalized": false,
+ "special": true
+ },
+ "100": {
+ "content": "<|channel>",
+ "single_word": false,
+ "lstrip": false,
+ "rstrip": false,
+ "normalized": false,
+ "special": true
+ },
+ "101": {
+ "content": "",
+ "single_word": false,
+ "lstrip": false,
+ "rstrip": false,
+ "normalized": false,
+ "special": true
+ },
+ "105": {
+ "content": "<|turn>",
+ "single_word": false,
+ "lstrip": false,
+ "rstrip": false,
+ "normalized": false,
+ "special": true
+ },
+ "106": {
+ "content": "",
+ "single_word": false,
+ "lstrip": false,
+ "rstrip": false,
+ "normalized": false,
+ "special": true
+ },
+ "255999": {
+ "content": "<|image>",
+ "single_word": false,
+ "lstrip": false,
+ "rstrip": false,
+ "normalized": false,
+ "special": true
+ },
+ "256000": {
+ "content": "<|audio>",
+ "single_word": false,
+ "lstrip": false,
+ "rstrip": false,
+ "normalized": false,
+ "special": true
+ },
+ "258880": {
+ "content": "<|image|>",
+ "single_word": false,
+ "lstrip": false,
+ "rstrip": false,
+ "normalized": false,
+ "special": true
+ },
+ "258881": {
+ "content": "<|audio|>",
+ "single_word": false,
+ "lstrip": false,
+ "rstrip": false,
+ "normalized": false,
+ "special": true
+ },
+ "258882": {
+ "content": "",
+ "single_word": false,
+ "lstrip": false,
+ "rstrip": false,
+ "normalized": false,
+ "special": true
+ },
+ "258883": {
+ "content": "",
+ "single_word": false,
+ "lstrip": false,
+ "rstrip": false,
+ "normalized": false,
+ "special": true
+ },
+ "258884": {
+ "content": "<|video|>",
+ "single_word": false,
+ "lstrip": false,
+ "rstrip": false,
+ "normalized": false,
+ "special": true
+ }
+ }
+}
diff --git a/checkpoint-180/trainer_state.json b/checkpoint-180/trainer_state.json
new file mode 100644
index 0000000000000000000000000000000000000000..76ed8c38fd4551d03f78f66ca97a3a293c3099c4
--- /dev/null
+++ b/checkpoint-180/trainer_state.json
@@ -0,0 +1,1834 @@
+{
+ "best_global_step": null,
+ "best_metric": null,
+ "best_model_checkpoint": null,
+ "epoch": 0.04260355029585799,
+ "eval_steps": 500,
+ "global_step": 180,
+ "is_hyper_param_search": false,
+ "is_local_process_zero": true,
+ "is_world_process_zero": true,
+ "log_history": [
+ {
+ "epoch": 0.00023668639053254438,
+ "grad_norm": NaN,
+ "learning_rate": 0.0,
+ "loss": 4.347962379455566,
+ "num_input_tokens_seen": 842,
+ "step": 1,
+ "train_runtime": 63.0289,
+ "train_tokens_per_second": 13.359
+ },
+ {
+ "epoch": 0.00047337278106508875,
+ "grad_norm": NaN,
+ "learning_rate": 0.0,
+ "loss": 4.634688854217529,
+ "num_input_tokens_seen": 1820,
+ "step": 2,
+ "train_runtime": 67.7033,
+ "train_tokens_per_second": 26.882
+ },
+ {
+ "epoch": 0.0007100591715976331,
+ "grad_norm": 13.944212913513184,
+ "learning_rate": 0.0,
+ "loss": 4.42241096496582,
+ "num_input_tokens_seen": 2754,
+ "step": 3,
+ "train_runtime": 72.324,
+ "train_tokens_per_second": 38.079
+ },
+ {
+ "epoch": 0.0009467455621301775,
+ "grad_norm": 13.445413589477539,
+ "learning_rate": 4e-05,
+ "loss": 4.682804107666016,
+ "num_input_tokens_seen": 3700,
+ "step": 4,
+ "train_runtime": 75.3925,
+ "train_tokens_per_second": 49.077
+ },
+ {
+ "epoch": 0.001183431952662722,
+ "grad_norm": 10.576727867126465,
+ "learning_rate": 8e-05,
+ "loss": 4.166159629821777,
+ "num_input_tokens_seen": 4714,
+ "step": 5,
+ "train_runtime": 78.474,
+ "train_tokens_per_second": 60.071
+ },
+ {
+ "epoch": 0.0014201183431952662,
+ "grad_norm": 8.077850341796875,
+ "learning_rate": 0.00012,
+ "loss": 3.2006261348724365,
+ "num_input_tokens_seen": 5674,
+ "step": 6,
+ "train_runtime": 81.6139,
+ "train_tokens_per_second": 69.522
+ },
+ {
+ "epoch": 0.0016568047337278107,
+ "grad_norm": 6.60789680480957,
+ "learning_rate": 0.00016,
+ "loss": 2.7067415714263916,
+ "num_input_tokens_seen": 6646,
+ "step": 7,
+ "train_runtime": 84.786,
+ "train_tokens_per_second": 78.386
+ },
+ {
+ "epoch": 0.001893491124260355,
+ "grad_norm": 9.175209045410156,
+ "learning_rate": 0.0002,
+ "loss": 2.367570638656616,
+ "num_input_tokens_seen": 7518,
+ "step": 8,
+ "train_runtime": 87.8178,
+ "train_tokens_per_second": 85.609
+ },
+ {
+ "epoch": 0.0021301775147928993,
+ "grad_norm": 3.438948631286621,
+ "learning_rate": 0.00019891891891891895,
+ "loss": 2.1013119220733643,
+ "num_input_tokens_seen": 8470,
+ "step": 9,
+ "train_runtime": 90.9458,
+ "train_tokens_per_second": 93.132
+ },
+ {
+ "epoch": 0.002366863905325444,
+ "grad_norm": 3.459472894668579,
+ "learning_rate": 0.00019783783783783784,
+ "loss": 1.8991940021514893,
+ "num_input_tokens_seen": 9458,
+ "step": 10,
+ "train_runtime": 94.0577,
+ "train_tokens_per_second": 100.555
+ },
+ {
+ "epoch": 0.0026035502958579883,
+ "grad_norm": 2.169750928878784,
+ "learning_rate": 0.00019675675675675678,
+ "loss": 1.7395055294036865,
+ "num_input_tokens_seen": 10456,
+ "step": 11,
+ "train_runtime": 97.2068,
+ "train_tokens_per_second": 107.565
+ },
+ {
+ "epoch": 0.0028402366863905324,
+ "grad_norm": 1.8638927936553955,
+ "learning_rate": 0.00019567567567567567,
+ "loss": 1.5622950792312622,
+ "num_input_tokens_seen": 11410,
+ "step": 12,
+ "train_runtime": 100.2911,
+ "train_tokens_per_second": 113.769
+ },
+ {
+ "epoch": 0.003076923076923077,
+ "grad_norm": 1.4741116762161255,
+ "learning_rate": 0.00019459459459459462,
+ "loss": 1.392699956893921,
+ "num_input_tokens_seen": 12316,
+ "step": 13,
+ "train_runtime": 103.3664,
+ "train_tokens_per_second": 119.149
+ },
+ {
+ "epoch": 0.0033136094674556214,
+ "grad_norm": 1.3238699436187744,
+ "learning_rate": 0.00019351351351351353,
+ "loss": 1.2535074949264526,
+ "num_input_tokens_seen": 13256,
+ "step": 14,
+ "train_runtime": 106.4031,
+ "train_tokens_per_second": 124.583
+ },
+ {
+ "epoch": 0.0035502958579881655,
+ "grad_norm": 1.276484489440918,
+ "learning_rate": 0.00019243243243243245,
+ "loss": 1.135048508644104,
+ "num_input_tokens_seen": 14228,
+ "step": 15,
+ "train_runtime": 109.4909,
+ "train_tokens_per_second": 129.947
+ },
+ {
+ "epoch": 0.00378698224852071,
+ "grad_norm": 1.1560865640640259,
+ "learning_rate": 0.00019135135135135137,
+ "loss": 1.0317097902297974,
+ "num_input_tokens_seen": 15192,
+ "step": 16,
+ "train_runtime": 112.5741,
+ "train_tokens_per_second": 134.951
+ },
+ {
+ "epoch": 0.004023668639053254,
+ "grad_norm": 1.1331353187561035,
+ "learning_rate": 0.00019027027027027028,
+ "loss": 0.9967219233512878,
+ "num_input_tokens_seen": 16130,
+ "step": 17,
+ "train_runtime": 115.6495,
+ "train_tokens_per_second": 139.473
+ },
+ {
+ "epoch": 0.004260355029585799,
+ "grad_norm": 1.1567702293395996,
+ "learning_rate": 0.0001891891891891892,
+ "loss": 0.905368447303772,
+ "num_input_tokens_seen": 17078,
+ "step": 18,
+ "train_runtime": 118.6796,
+ "train_tokens_per_second": 143.9
+ },
+ {
+ "epoch": 0.004497041420118343,
+ "grad_norm": 0.9157600998878479,
+ "learning_rate": 0.00018810810810810814,
+ "loss": 0.8075690865516663,
+ "num_input_tokens_seen": 18062,
+ "step": 19,
+ "train_runtime": 121.8602,
+ "train_tokens_per_second": 148.219
+ },
+ {
+ "epoch": 0.004733727810650888,
+ "grad_norm": 0.8306605815887451,
+ "learning_rate": 0.00018702702702702703,
+ "loss": 0.8007175922393799,
+ "num_input_tokens_seen": 19078,
+ "step": 20,
+ "train_runtime": 125.0019,
+ "train_tokens_per_second": 152.622
+ },
+ {
+ "epoch": 0.004970414201183432,
+ "grad_norm": 0.690018355846405,
+ "learning_rate": 0.00018594594594594597,
+ "loss": 0.7810791730880737,
+ "num_input_tokens_seen": 20068,
+ "step": 21,
+ "train_runtime": 128.0932,
+ "train_tokens_per_second": 156.667
+ },
+ {
+ "epoch": 0.005207100591715977,
+ "grad_norm": 1.4768867492675781,
+ "learning_rate": 0.00018486486486486486,
+ "loss": 0.7172120809555054,
+ "num_input_tokens_seen": 20972,
+ "step": 22,
+ "train_runtime": 131.0901,
+ "train_tokens_per_second": 159.982
+ },
+ {
+ "epoch": 0.00544378698224852,
+ "grad_norm": 0.7999809980392456,
+ "learning_rate": 0.0001837837837837838,
+ "loss": 0.6333726048469543,
+ "num_input_tokens_seen": 21900,
+ "step": 23,
+ "train_runtime": 134.212,
+ "train_tokens_per_second": 163.175
+ },
+ {
+ "epoch": 0.005680473372781065,
+ "grad_norm": 0.9285996556282043,
+ "learning_rate": 0.00018270270270270272,
+ "loss": 0.6750729084014893,
+ "num_input_tokens_seen": 22770,
+ "step": 24,
+ "train_runtime": 137.3193,
+ "train_tokens_per_second": 165.818
+ },
+ {
+ "epoch": 0.005917159763313609,
+ "grad_norm": 0.7733820080757141,
+ "learning_rate": 0.00018162162162162164,
+ "loss": 0.672032356262207,
+ "num_input_tokens_seen": 23746,
+ "step": 25,
+ "train_runtime": 140.3175,
+ "train_tokens_per_second": 169.23
+ },
+ {
+ "epoch": 0.006153846153846154,
+ "grad_norm": 0.73944091796875,
+ "learning_rate": 0.00018054054054054055,
+ "loss": 0.6716101169586182,
+ "num_input_tokens_seen": 24716,
+ "step": 26,
+ "train_runtime": 143.3683,
+ "train_tokens_per_second": 172.395
+ },
+ {
+ "epoch": 0.006390532544378698,
+ "grad_norm": 0.7659372091293335,
+ "learning_rate": 0.00017945945945945947,
+ "loss": 0.6067175269126892,
+ "num_input_tokens_seen": 25724,
+ "step": 27,
+ "train_runtime": 146.5377,
+ "train_tokens_per_second": 175.545
+ },
+ {
+ "epoch": 0.006627218934911243,
+ "grad_norm": 0.5428617000579834,
+ "learning_rate": 0.00017837837837837839,
+ "loss": 0.6083395481109619,
+ "num_input_tokens_seen": 26712,
+ "step": 28,
+ "train_runtime": 149.6429,
+ "train_tokens_per_second": 178.505
+ },
+ {
+ "epoch": 0.006863905325443787,
+ "grad_norm": 0.5915114283561707,
+ "learning_rate": 0.00017729729729729733,
+ "loss": 0.6463742256164551,
+ "num_input_tokens_seen": 27704,
+ "step": 29,
+ "train_runtime": 152.7501,
+ "train_tokens_per_second": 181.368
+ },
+ {
+ "epoch": 0.007100591715976331,
+ "grad_norm": 1.3350757360458374,
+ "learning_rate": 0.00017621621621621622,
+ "loss": 0.5705698132514954,
+ "num_input_tokens_seen": 28648,
+ "step": 30,
+ "train_runtime": 155.8918,
+ "train_tokens_per_second": 183.768
+ },
+ {
+ "epoch": 0.0073372781065088755,
+ "grad_norm": 0.68421471118927,
+ "learning_rate": 0.00017513513513513516,
+ "loss": 0.5660746097564697,
+ "num_input_tokens_seen": 29622,
+ "step": 31,
+ "train_runtime": 163.3209,
+ "train_tokens_per_second": 181.373
+ },
+ {
+ "epoch": 0.00757396449704142,
+ "grad_norm": 0.5762604475021362,
+ "learning_rate": 0.00017405405405405405,
+ "loss": 0.5483309030532837,
+ "num_input_tokens_seen": 30534,
+ "step": 32,
+ "train_runtime": 166.5185,
+ "train_tokens_per_second": 183.367
+ },
+ {
+ "epoch": 0.0078106508875739646,
+ "grad_norm": 0.5751784443855286,
+ "learning_rate": 0.000172972972972973,
+ "loss": 0.5207005739212036,
+ "num_input_tokens_seen": 31458,
+ "step": 33,
+ "train_runtime": 169.7055,
+ "train_tokens_per_second": 185.368
+ },
+ {
+ "epoch": 0.008047337278106508,
+ "grad_norm": 0.5752039551734924,
+ "learning_rate": 0.0001718918918918919,
+ "loss": 0.5534584522247314,
+ "num_input_tokens_seen": 32424,
+ "step": 34,
+ "train_runtime": 172.9184,
+ "train_tokens_per_second": 187.51
+ },
+ {
+ "epoch": 0.008284023668639054,
+ "grad_norm": 0.5918363332748413,
+ "learning_rate": 0.00017081081081081083,
+ "loss": 0.482683002948761,
+ "num_input_tokens_seen": 33398,
+ "step": 35,
+ "train_runtime": 176.2158,
+ "train_tokens_per_second": 189.529
+ },
+ {
+ "epoch": 0.008520710059171597,
+ "grad_norm": 0.6921846270561218,
+ "learning_rate": 0.00016972972972972974,
+ "loss": 0.46112507581710815,
+ "num_input_tokens_seen": 34324,
+ "step": 36,
+ "train_runtime": 179.3173,
+ "train_tokens_per_second": 191.415
+ },
+ {
+ "epoch": 0.008757396449704143,
+ "grad_norm": 0.6429721117019653,
+ "learning_rate": 0.00016864864864864866,
+ "loss": 0.45629197359085083,
+ "num_input_tokens_seen": 35270,
+ "step": 37,
+ "train_runtime": 182.3558,
+ "train_tokens_per_second": 193.413
+ },
+ {
+ "epoch": 0.008994082840236686,
+ "grad_norm": 0.5477020740509033,
+ "learning_rate": 0.00016756756756756757,
+ "loss": 0.4547877907752991,
+ "num_input_tokens_seen": 36206,
+ "step": 38,
+ "train_runtime": 185.4483,
+ "train_tokens_per_second": 195.235
+ },
+ {
+ "epoch": 0.009230769230769232,
+ "grad_norm": 0.6158634424209595,
+ "learning_rate": 0.00016648648648648652,
+ "loss": 0.367482990026474,
+ "num_input_tokens_seen": 37080,
+ "step": 39,
+ "train_runtime": 188.5114,
+ "train_tokens_per_second": 196.699
+ },
+ {
+ "epoch": 0.009467455621301775,
+ "grad_norm": 0.5405091643333435,
+ "learning_rate": 0.0001654054054054054,
+ "loss": 0.42049410939216614,
+ "num_input_tokens_seen": 38030,
+ "step": 40,
+ "train_runtime": 191.5986,
+ "train_tokens_per_second": 198.488
+ },
+ {
+ "epoch": 0.009704142011834319,
+ "grad_norm": 0.6389183402061462,
+ "learning_rate": 0.00016432432432432435,
+ "loss": 0.3935014307498932,
+ "num_input_tokens_seen": 38942,
+ "step": 41,
+ "train_runtime": 194.8426,
+ "train_tokens_per_second": 199.864
+ },
+ {
+ "epoch": 0.009940828402366864,
+ "grad_norm": 0.6742956042289734,
+ "learning_rate": 0.00016324324324324324,
+ "loss": 0.4373692274093628,
+ "num_input_tokens_seen": 39902,
+ "step": 42,
+ "train_runtime": 197.8938,
+ "train_tokens_per_second": 201.633
+ },
+ {
+ "epoch": 0.010177514792899408,
+ "grad_norm": 0.6083794236183167,
+ "learning_rate": 0.00016216216216216218,
+ "loss": 0.3776765465736389,
+ "num_input_tokens_seen": 40858,
+ "step": 43,
+ "train_runtime": 200.9708,
+ "train_tokens_per_second": 203.303
+ },
+ {
+ "epoch": 0.010414201183431953,
+ "grad_norm": 0.5469992160797119,
+ "learning_rate": 0.0001610810810810811,
+ "loss": 0.31133151054382324,
+ "num_input_tokens_seen": 41760,
+ "step": 44,
+ "train_runtime": 204.1304,
+ "train_tokens_per_second": 204.575
+ },
+ {
+ "epoch": 0.010650887573964497,
+ "grad_norm": 0.6452498435974121,
+ "learning_rate": 0.00016,
+ "loss": 0.42721521854400635,
+ "num_input_tokens_seen": 42740,
+ "step": 45,
+ "train_runtime": 207.1845,
+ "train_tokens_per_second": 206.29
+ },
+ {
+ "epoch": 0.01088757396449704,
+ "grad_norm": 0.5988945960998535,
+ "learning_rate": 0.00015891891891891893,
+ "loss": 0.41375258564949036,
+ "num_input_tokens_seen": 43750,
+ "step": 46,
+ "train_runtime": 210.2912,
+ "train_tokens_per_second": 208.045
+ },
+ {
+ "epoch": 0.011124260355029586,
+ "grad_norm": 0.6452038288116455,
+ "learning_rate": 0.00015783783783783785,
+ "loss": 0.3485276699066162,
+ "num_input_tokens_seen": 44690,
+ "step": 47,
+ "train_runtime": 213.3684,
+ "train_tokens_per_second": 209.45
+ },
+ {
+ "epoch": 0.01136094674556213,
+ "grad_norm": 0.7041454911231995,
+ "learning_rate": 0.00015675675675675676,
+ "loss": 0.3296637535095215,
+ "num_input_tokens_seen": 45666,
+ "step": 48,
+ "train_runtime": 216.4459,
+ "train_tokens_per_second": 210.981
+ },
+ {
+ "epoch": 0.011597633136094675,
+ "grad_norm": 0.6726324558258057,
+ "learning_rate": 0.0001556756756756757,
+ "loss": 0.3179457187652588,
+ "num_input_tokens_seen": 46560,
+ "step": 49,
+ "train_runtime": 219.4697,
+ "train_tokens_per_second": 212.148
+ },
+ {
+ "epoch": 0.011834319526627219,
+ "grad_norm": 0.7513703107833862,
+ "learning_rate": 0.0001545945945945946,
+ "loss": 0.3722391426563263,
+ "num_input_tokens_seen": 47528,
+ "step": 50,
+ "train_runtime": 222.596,
+ "train_tokens_per_second": 213.517
+ },
+ {
+ "epoch": 0.012071005917159764,
+ "grad_norm": 0.6454160213470459,
+ "learning_rate": 0.00015351351351351354,
+ "loss": 0.283376544713974,
+ "num_input_tokens_seen": 48468,
+ "step": 51,
+ "train_runtime": 225.7262,
+ "train_tokens_per_second": 214.72
+ },
+ {
+ "epoch": 0.012307692307692308,
+ "grad_norm": 0.9116244316101074,
+ "learning_rate": 0.00015243243243243243,
+ "loss": 0.4286752939224243,
+ "num_input_tokens_seen": 49444,
+ "step": 52,
+ "train_runtime": 228.803,
+ "train_tokens_per_second": 216.099
+ },
+ {
+ "epoch": 0.012544378698224851,
+ "grad_norm": 0.6157099604606628,
+ "learning_rate": 0.00015135135135135137,
+ "loss": 0.27355721592903137,
+ "num_input_tokens_seen": 50424,
+ "step": 53,
+ "train_runtime": 231.8413,
+ "train_tokens_per_second": 217.494
+ },
+ {
+ "epoch": 0.012781065088757397,
+ "grad_norm": 0.6335808038711548,
+ "learning_rate": 0.00015027027027027028,
+ "loss": 0.37197017669677734,
+ "num_input_tokens_seen": 51328,
+ "step": 54,
+ "train_runtime": 234.9265,
+ "train_tokens_per_second": 218.485
+ },
+ {
+ "epoch": 0.01301775147928994,
+ "grad_norm": 0.6838424205780029,
+ "learning_rate": 0.0001491891891891892,
+ "loss": 0.3504602909088135,
+ "num_input_tokens_seen": 52274,
+ "step": 55,
+ "train_runtime": 238.0114,
+ "train_tokens_per_second": 219.628
+ },
+ {
+ "epoch": 0.013254437869822486,
+ "grad_norm": 0.6566418409347534,
+ "learning_rate": 0.00014810810810810812,
+ "loss": 0.3189927935600281,
+ "num_input_tokens_seen": 53214,
+ "step": 56,
+ "train_runtime": 241.0892,
+ "train_tokens_per_second": 220.723
+ },
+ {
+ "epoch": 0.01349112426035503,
+ "grad_norm": 0.6112971901893616,
+ "learning_rate": 0.00014702702702702703,
+ "loss": 0.3080369830131531,
+ "num_input_tokens_seen": 54162,
+ "step": 57,
+ "train_runtime": 244.2461,
+ "train_tokens_per_second": 221.752
+ },
+ {
+ "epoch": 0.013727810650887575,
+ "grad_norm": 0.6469757556915283,
+ "learning_rate": 0.00014594594594594595,
+ "loss": 0.24279898405075073,
+ "num_input_tokens_seen": 55158,
+ "step": 58,
+ "train_runtime": 247.4064,
+ "train_tokens_per_second": 222.945
+ },
+ {
+ "epoch": 0.013964497041420118,
+ "grad_norm": 0.7099384665489197,
+ "learning_rate": 0.0001448648648648649,
+ "loss": 0.31188705563545227,
+ "num_input_tokens_seen": 56096,
+ "step": 59,
+ "train_runtime": 250.4708,
+ "train_tokens_per_second": 223.962
+ },
+ {
+ "epoch": 0.014201183431952662,
+ "grad_norm": 0.5927040576934814,
+ "learning_rate": 0.00014378378378378378,
+ "loss": 0.25696688890457153,
+ "num_input_tokens_seen": 57012,
+ "step": 60,
+ "train_runtime": 253.7131,
+ "train_tokens_per_second": 224.711
+ },
+ {
+ "epoch": 0.014437869822485207,
+ "grad_norm": 0.5483338236808777,
+ "learning_rate": 0.00014270270270270272,
+ "loss": 0.2725352644920349,
+ "num_input_tokens_seen": 58036,
+ "step": 61,
+ "train_runtime": 260.2844,
+ "train_tokens_per_second": 222.971
+ },
+ {
+ "epoch": 0.014674556213017751,
+ "grad_norm": 0.6806847453117371,
+ "learning_rate": 0.00014162162162162161,
+ "loss": 0.31959983706474304,
+ "num_input_tokens_seen": 59026,
+ "step": 62,
+ "train_runtime": 263.3802,
+ "train_tokens_per_second": 224.109
+ },
+ {
+ "epoch": 0.014911242603550296,
+ "grad_norm": 0.6916660070419312,
+ "learning_rate": 0.00014054054054054056,
+ "loss": 0.312305748462677,
+ "num_input_tokens_seen": 59982,
+ "step": 63,
+ "train_runtime": 266.3794,
+ "train_tokens_per_second": 225.175
+ },
+ {
+ "epoch": 0.01514792899408284,
+ "grad_norm": 0.5516266226768494,
+ "learning_rate": 0.00013945945945945947,
+ "loss": 0.2653953433036804,
+ "num_input_tokens_seen": 60926,
+ "step": 64,
+ "train_runtime": 269.3893,
+ "train_tokens_per_second": 226.163
+ },
+ {
+ "epoch": 0.015384615384615385,
+ "grad_norm": 0.5579270720481873,
+ "learning_rate": 0.0001383783783783784,
+ "loss": 0.20462818443775177,
+ "num_input_tokens_seen": 61844,
+ "step": 65,
+ "train_runtime": 272.3982,
+ "train_tokens_per_second": 227.035
+ },
+ {
+ "epoch": 0.015621301775147929,
+ "grad_norm": 0.7645660638809204,
+ "learning_rate": 0.0001372972972972973,
+ "loss": 0.2650893032550812,
+ "num_input_tokens_seen": 62766,
+ "step": 66,
+ "train_runtime": 275.404,
+ "train_tokens_per_second": 227.905
+ },
+ {
+ "epoch": 0.015857988165680473,
+ "grad_norm": 0.5974738001823425,
+ "learning_rate": 0.00013621621621621622,
+ "loss": 0.21247778832912445,
+ "num_input_tokens_seen": 63690,
+ "step": 67,
+ "train_runtime": 278.4021,
+ "train_tokens_per_second": 228.77
+ },
+ {
+ "epoch": 0.016094674556213016,
+ "grad_norm": 0.5831193327903748,
+ "learning_rate": 0.00013513513513513514,
+ "loss": 0.2024172842502594,
+ "num_input_tokens_seen": 64496,
+ "step": 68,
+ "train_runtime": 281.4313,
+ "train_tokens_per_second": 229.171
+ },
+ {
+ "epoch": 0.016331360946745564,
+ "grad_norm": 0.8678795695304871,
+ "learning_rate": 0.00013405405405405408,
+ "loss": 0.31490638852119446,
+ "num_input_tokens_seen": 65460,
+ "step": 69,
+ "train_runtime": 284.6137,
+ "train_tokens_per_second": 229.996
+ },
+ {
+ "epoch": 0.016568047337278107,
+ "grad_norm": 0.7611344456672668,
+ "learning_rate": 0.00013297297297297297,
+ "loss": 0.24661925435066223,
+ "num_input_tokens_seen": 66374,
+ "step": 70,
+ "train_runtime": 287.7742,
+ "train_tokens_per_second": 230.646
+ },
+ {
+ "epoch": 0.01680473372781065,
+ "grad_norm": 0.6685308814048767,
+ "learning_rate": 0.0001318918918918919,
+ "loss": 0.1843535602092743,
+ "num_input_tokens_seen": 67112,
+ "step": 71,
+ "train_runtime": 290.9321,
+ "train_tokens_per_second": 230.679
+ },
+ {
+ "epoch": 0.017041420118343194,
+ "grad_norm": 0.6150883436203003,
+ "learning_rate": 0.0001308108108108108,
+ "loss": 0.23372295498847961,
+ "num_input_tokens_seen": 68154,
+ "step": 72,
+ "train_runtime": 294.0809,
+ "train_tokens_per_second": 231.753
+ },
+ {
+ "epoch": 0.017278106508875738,
+ "grad_norm": 0.6938812732696533,
+ "learning_rate": 0.00012972972972972974,
+ "loss": 0.22619186341762543,
+ "num_input_tokens_seen": 69102,
+ "step": 73,
+ "train_runtime": 297.1843,
+ "train_tokens_per_second": 232.522
+ },
+ {
+ "epoch": 0.017514792899408285,
+ "grad_norm": 0.617787778377533,
+ "learning_rate": 0.00012864864864864866,
+ "loss": 0.2535328269004822,
+ "num_input_tokens_seen": 70010,
+ "step": 74,
+ "train_runtime": 300.2851,
+ "train_tokens_per_second": 233.145
+ },
+ {
+ "epoch": 0.01775147928994083,
+ "grad_norm": 0.6884900331497192,
+ "learning_rate": 0.00012756756756756758,
+ "loss": 0.21632739901542664,
+ "num_input_tokens_seen": 70940,
+ "step": 75,
+ "train_runtime": 303.3533,
+ "train_tokens_per_second": 233.853
+ },
+ {
+ "epoch": 0.017988165680473372,
+ "grad_norm": 0.5782189965248108,
+ "learning_rate": 0.0001264864864864865,
+ "loss": 0.20167675614356995,
+ "num_input_tokens_seen": 71862,
+ "step": 76,
+ "train_runtime": 306.4503,
+ "train_tokens_per_second": 234.498
+ },
+ {
+ "epoch": 0.018224852071005916,
+ "grad_norm": 0.6541483402252197,
+ "learning_rate": 0.0001254054054054054,
+ "loss": 0.23231221735477448,
+ "num_input_tokens_seen": 72812,
+ "step": 77,
+ "train_runtime": 309.5743,
+ "train_tokens_per_second": 235.2
+ },
+ {
+ "epoch": 0.018461538461538463,
+ "grad_norm": 0.7524886727333069,
+ "learning_rate": 0.00012432432432432433,
+ "loss": 0.27632662653923035,
+ "num_input_tokens_seen": 73774,
+ "step": 78,
+ "train_runtime": 312.6963,
+ "train_tokens_per_second": 235.929
+ },
+ {
+ "epoch": 0.018698224852071007,
+ "grad_norm": 0.716282308101654,
+ "learning_rate": 0.00012324324324324327,
+ "loss": 0.2734939754009247,
+ "num_input_tokens_seen": 74716,
+ "step": 79,
+ "train_runtime": 315.7612,
+ "train_tokens_per_second": 236.622
+ },
+ {
+ "epoch": 0.01893491124260355,
+ "grad_norm": 0.7010631561279297,
+ "learning_rate": 0.00012216216216216216,
+ "loss": 0.2774362564086914,
+ "num_input_tokens_seen": 75692,
+ "step": 80,
+ "train_runtime": 318.7983,
+ "train_tokens_per_second": 237.429
+ },
+ {
+ "epoch": 0.019171597633136094,
+ "grad_norm": 0.6561240553855896,
+ "learning_rate": 0.00012108108108108109,
+ "loss": 0.21640686690807343,
+ "num_input_tokens_seen": 76558,
+ "step": 81,
+ "train_runtime": 321.889,
+ "train_tokens_per_second": 237.84
+ },
+ {
+ "epoch": 0.019408284023668638,
+ "grad_norm": 0.5552831888198853,
+ "learning_rate": 0.00012,
+ "loss": 0.21232956647872925,
+ "num_input_tokens_seen": 77408,
+ "step": 82,
+ "train_runtime": 325.0469,
+ "train_tokens_per_second": 238.144
+ },
+ {
+ "epoch": 0.019644970414201185,
+ "grad_norm": 0.5905120372772217,
+ "learning_rate": 0.00011891891891891893,
+ "loss": 0.22443550825119019,
+ "num_input_tokens_seen": 78278,
+ "step": 83,
+ "train_runtime": 328.099,
+ "train_tokens_per_second": 238.58
+ },
+ {
+ "epoch": 0.01988165680473373,
+ "grad_norm": 0.6606366634368896,
+ "learning_rate": 0.00011783783783783784,
+ "loss": 0.2400747388601303,
+ "num_input_tokens_seen": 79262,
+ "step": 84,
+ "train_runtime": 331.1361,
+ "train_tokens_per_second": 239.364
+ },
+ {
+ "epoch": 0.020118343195266272,
+ "grad_norm": 0.6508055329322815,
+ "learning_rate": 0.00011675675675675676,
+ "loss": 0.19072499871253967,
+ "num_input_tokens_seen": 80144,
+ "step": 85,
+ "train_runtime": 334.2158,
+ "train_tokens_per_second": 239.797
+ },
+ {
+ "epoch": 0.020355029585798816,
+ "grad_norm": 0.606562614440918,
+ "learning_rate": 0.00011567567567567568,
+ "loss": 0.20951862633228302,
+ "num_input_tokens_seen": 81108,
+ "step": 86,
+ "train_runtime": 337.3105,
+ "train_tokens_per_second": 240.455
+ },
+ {
+ "epoch": 0.02059171597633136,
+ "grad_norm": 0.6757521629333496,
+ "learning_rate": 0.00011459459459459461,
+ "loss": 0.17614495754241943,
+ "num_input_tokens_seen": 82032,
+ "step": 87,
+ "train_runtime": 340.3438,
+ "train_tokens_per_second": 241.027
+ },
+ {
+ "epoch": 0.020828402366863907,
+ "grad_norm": 0.671481728553772,
+ "learning_rate": 0.00011351351351351351,
+ "loss": 0.1901710331439972,
+ "num_input_tokens_seen": 82954,
+ "step": 88,
+ "train_runtime": 343.4059,
+ "train_tokens_per_second": 241.563
+ },
+ {
+ "epoch": 0.02106508875739645,
+ "grad_norm": 0.5919027924537659,
+ "learning_rate": 0.00011243243243243244,
+ "loss": 0.1563737988471985,
+ "num_input_tokens_seen": 83880,
+ "step": 89,
+ "train_runtime": 346.5125,
+ "train_tokens_per_second": 242.069
+ },
+ {
+ "epoch": 0.021301775147928994,
+ "grad_norm": 0.7108309268951416,
+ "learning_rate": 0.00011135135135135135,
+ "loss": 0.17256894707679749,
+ "num_input_tokens_seen": 84852,
+ "step": 90,
+ "train_runtime": 349.6508,
+ "train_tokens_per_second": 242.676
+ },
+ {
+ "epoch": 0.021538461538461538,
+ "grad_norm": 0.8608420491218567,
+ "learning_rate": 0.00011027027027027029,
+ "loss": 0.1845530867576599,
+ "num_input_tokens_seen": 85820,
+ "step": 91,
+ "train_runtime": 356.267,
+ "train_tokens_per_second": 240.887
+ },
+ {
+ "epoch": 0.02177514792899408,
+ "grad_norm": 0.9677010774612427,
+ "learning_rate": 0.00010918918918918919,
+ "loss": 0.24358105659484863,
+ "num_input_tokens_seen": 86770,
+ "step": 92,
+ "train_runtime": 359.3722,
+ "train_tokens_per_second": 241.449
+ },
+ {
+ "epoch": 0.02201183431952663,
+ "grad_norm": 0.6302966475486755,
+ "learning_rate": 0.00010810810810810812,
+ "loss": 0.18155133724212646,
+ "num_input_tokens_seen": 87736,
+ "step": 93,
+ "train_runtime": 362.5874,
+ "train_tokens_per_second": 241.972
+ },
+ {
+ "epoch": 0.022248520710059172,
+ "grad_norm": 0.7171812653541565,
+ "learning_rate": 0.00010702702702702702,
+ "loss": 0.15349116921424866,
+ "num_input_tokens_seen": 88626,
+ "step": 94,
+ "train_runtime": 365.7055,
+ "train_tokens_per_second": 242.343
+ },
+ {
+ "epoch": 0.022485207100591716,
+ "grad_norm": 0.7326887845993042,
+ "learning_rate": 0.00010594594594594595,
+ "loss": 0.18243758380413055,
+ "num_input_tokens_seen": 89548,
+ "step": 95,
+ "train_runtime": 368.799,
+ "train_tokens_per_second": 242.81
+ },
+ {
+ "epoch": 0.02272189349112426,
+ "grad_norm": 0.6686019897460938,
+ "learning_rate": 0.00010486486486486487,
+ "loss": 0.15622465312480927,
+ "num_input_tokens_seen": 90504,
+ "step": 96,
+ "train_runtime": 371.8404,
+ "train_tokens_per_second": 243.395
+ },
+ {
+ "epoch": 0.022958579881656806,
+ "grad_norm": 0.6306503415107727,
+ "learning_rate": 0.0001037837837837838,
+ "loss": 0.1272117793560028,
+ "num_input_tokens_seen": 91454,
+ "step": 97,
+ "train_runtime": 375.0199,
+ "train_tokens_per_second": 243.864
+ },
+ {
+ "epoch": 0.02319526627218935,
+ "grad_norm": 0.6068493127822876,
+ "learning_rate": 0.0001027027027027027,
+ "loss": 0.1577155441045761,
+ "num_input_tokens_seen": 92466,
+ "step": 98,
+ "train_runtime": 378.0978,
+ "train_tokens_per_second": 244.556
+ },
+ {
+ "epoch": 0.023431952662721894,
+ "grad_norm": 0.7263635993003845,
+ "learning_rate": 0.00010162162162162163,
+ "loss": 0.1845340132713318,
+ "num_input_tokens_seen": 93398,
+ "step": 99,
+ "train_runtime": 381.1358,
+ "train_tokens_per_second": 245.052
+ },
+ {
+ "epoch": 0.023668639053254437,
+ "grad_norm": 0.6466794013977051,
+ "learning_rate": 0.00010054054054054053,
+ "loss": 0.14443959295749664,
+ "num_input_tokens_seen": 94230,
+ "step": 100,
+ "train_runtime": 384.2906,
+ "train_tokens_per_second": 245.205
+ },
+ {
+ "epoch": 0.02390532544378698,
+ "grad_norm": 0.6924043297767639,
+ "learning_rate": 9.945945945945948e-05,
+ "loss": 0.18473166227340698,
+ "num_input_tokens_seen": 95132,
+ "step": 101,
+ "train_runtime": 387.5005,
+ "train_tokens_per_second": 245.502
+ },
+ {
+ "epoch": 0.024142011834319528,
+ "grad_norm": 0.732337236404419,
+ "learning_rate": 9.837837837837839e-05,
+ "loss": 0.16353662312030792,
+ "num_input_tokens_seen": 96080,
+ "step": 102,
+ "train_runtime": 390.584,
+ "train_tokens_per_second": 245.991
+ },
+ {
+ "epoch": 0.02437869822485207,
+ "grad_norm": 0.5956932902336121,
+ "learning_rate": 9.729729729729731e-05,
+ "loss": 0.16359847784042358,
+ "num_input_tokens_seen": 97048,
+ "step": 103,
+ "train_runtime": 393.6474,
+ "train_tokens_per_second": 246.535
+ },
+ {
+ "epoch": 0.024615384615384615,
+ "grad_norm": 0.6137388348579407,
+ "learning_rate": 9.621621621621622e-05,
+ "loss": 0.15779417753219604,
+ "num_input_tokens_seen": 97986,
+ "step": 104,
+ "train_runtime": 396.7042,
+ "train_tokens_per_second": 247.0
+ },
+ {
+ "epoch": 0.02485207100591716,
+ "grad_norm": 0.7936599254608154,
+ "learning_rate": 9.513513513513514e-05,
+ "loss": 0.21357379853725433,
+ "num_input_tokens_seen": 98894,
+ "step": 105,
+ "train_runtime": 399.8228,
+ "train_tokens_per_second": 247.345
+ },
+ {
+ "epoch": 0.025088757396449703,
+ "grad_norm": 0.6574507355690002,
+ "learning_rate": 9.405405405405407e-05,
+ "loss": 0.1685573309659958,
+ "num_input_tokens_seen": 99856,
+ "step": 106,
+ "train_runtime": 402.8867,
+ "train_tokens_per_second": 247.851
+ },
+ {
+ "epoch": 0.02532544378698225,
+ "grad_norm": 0.5891634821891785,
+ "learning_rate": 9.297297297297299e-05,
+ "loss": 0.12082140147686005,
+ "num_input_tokens_seen": 100810,
+ "step": 107,
+ "train_runtime": 405.9752,
+ "train_tokens_per_second": 248.316
+ },
+ {
+ "epoch": 0.025562130177514793,
+ "grad_norm": 0.6624342799186707,
+ "learning_rate": 9.18918918918919e-05,
+ "loss": 0.1473795771598816,
+ "num_input_tokens_seen": 101744,
+ "step": 108,
+ "train_runtime": 409.0084,
+ "train_tokens_per_second": 248.758
+ },
+ {
+ "epoch": 0.025798816568047337,
+ "grad_norm": 0.5968036651611328,
+ "learning_rate": 9.081081081081082e-05,
+ "loss": 0.12923751771450043,
+ "num_input_tokens_seen": 102702,
+ "step": 109,
+ "train_runtime": 412.199,
+ "train_tokens_per_second": 249.156
+ },
+ {
+ "epoch": 0.02603550295857988,
+ "grad_norm": 0.5954580307006836,
+ "learning_rate": 8.972972972972973e-05,
+ "loss": 0.1161399856209755,
+ "num_input_tokens_seen": 103672,
+ "step": 110,
+ "train_runtime": 415.3736,
+ "train_tokens_per_second": 249.587
+ },
+ {
+ "epoch": 0.026272189349112424,
+ "grad_norm": 0.6139379739761353,
+ "learning_rate": 8.864864864864866e-05,
+ "loss": 0.12352899461984634,
+ "num_input_tokens_seen": 104598,
+ "step": 111,
+ "train_runtime": 418.3271,
+ "train_tokens_per_second": 250.039
+ },
+ {
+ "epoch": 0.02650887573964497,
+ "grad_norm": 0.7933095097541809,
+ "learning_rate": 8.756756756756758e-05,
+ "loss": 0.1709994524717331,
+ "num_input_tokens_seen": 105580,
+ "step": 112,
+ "train_runtime": 421.3565,
+ "train_tokens_per_second": 250.572
+ },
+ {
+ "epoch": 0.026745562130177515,
+ "grad_norm": 0.7641370892524719,
+ "learning_rate": 8.64864864864865e-05,
+ "loss": 0.13086311519145966,
+ "num_input_tokens_seen": 106444,
+ "step": 113,
+ "train_runtime": 424.4668,
+ "train_tokens_per_second": 250.771
+ },
+ {
+ "epoch": 0.02698224852071006,
+ "grad_norm": 0.7611938118934631,
+ "learning_rate": 8.540540540540541e-05,
+ "loss": 0.16537640988826752,
+ "num_input_tokens_seen": 107454,
+ "step": 114,
+ "train_runtime": 427.5,
+ "train_tokens_per_second": 251.354
+ },
+ {
+ "epoch": 0.027218934911242602,
+ "grad_norm": 0.7495304346084595,
+ "learning_rate": 8.432432432432433e-05,
+ "loss": 0.14228317141532898,
+ "num_input_tokens_seen": 108426,
+ "step": 115,
+ "train_runtime": 430.5035,
+ "train_tokens_per_second": 251.859
+ },
+ {
+ "epoch": 0.02745562130177515,
+ "grad_norm": 1.2078282833099365,
+ "learning_rate": 8.324324324324326e-05,
+ "loss": 0.10530930012464523,
+ "num_input_tokens_seen": 109446,
+ "step": 116,
+ "train_runtime": 433.4964,
+ "train_tokens_per_second": 252.473
+ },
+ {
+ "epoch": 0.027692307692307693,
+ "grad_norm": 0.6765572428703308,
+ "learning_rate": 8.216216216216217e-05,
+ "loss": 0.10717278718948364,
+ "num_input_tokens_seen": 110374,
+ "step": 117,
+ "train_runtime": 436.5901,
+ "train_tokens_per_second": 252.809
+ },
+ {
+ "epoch": 0.027928994082840237,
+ "grad_norm": 0.8501551151275635,
+ "learning_rate": 8.108108108108109e-05,
+ "loss": 0.17784273624420166,
+ "num_input_tokens_seen": 111340,
+ "step": 118,
+ "train_runtime": 439.6526,
+ "train_tokens_per_second": 253.245
+ },
+ {
+ "epoch": 0.02816568047337278,
+ "grad_norm": 0.7583926320075989,
+ "learning_rate": 8e-05,
+ "loss": 0.16588062047958374,
+ "num_input_tokens_seen": 112288,
+ "step": 119,
+ "train_runtime": 442.8046,
+ "train_tokens_per_second": 253.584
+ },
+ {
+ "epoch": 0.028402366863905324,
+ "grad_norm": 0.6620173454284668,
+ "learning_rate": 7.891891891891892e-05,
+ "loss": 0.11676295101642609,
+ "num_input_tokens_seen": 113206,
+ "step": 120,
+ "train_runtime": 445.9682,
+ "train_tokens_per_second": 253.843
+ },
+ {
+ "epoch": 0.02863905325443787,
+ "grad_norm": 0.6844660639762878,
+ "learning_rate": 7.783783783783785e-05,
+ "loss": 0.09329869598150253,
+ "num_input_tokens_seen": 114134,
+ "step": 121,
+ "train_runtime": 452.5692,
+ "train_tokens_per_second": 252.191
+ },
+ {
+ "epoch": 0.028875739644970415,
+ "grad_norm": 0.7968376874923706,
+ "learning_rate": 7.675675675675677e-05,
+ "loss": 0.13320565223693848,
+ "num_input_tokens_seen": 115088,
+ "step": 122,
+ "train_runtime": 455.6245,
+ "train_tokens_per_second": 252.594
+ },
+ {
+ "epoch": 0.02911242603550296,
+ "grad_norm": 0.8079301118850708,
+ "learning_rate": 7.567567567567568e-05,
+ "loss": 0.1387016326189041,
+ "num_input_tokens_seen": 116006,
+ "step": 123,
+ "train_runtime": 458.7601,
+ "train_tokens_per_second": 252.869
+ },
+ {
+ "epoch": 0.029349112426035502,
+ "grad_norm": 1.1301748752593994,
+ "learning_rate": 7.45945945945946e-05,
+ "loss": 0.197790265083313,
+ "num_input_tokens_seen": 116958,
+ "step": 124,
+ "train_runtime": 461.861,
+ "train_tokens_per_second": 253.232
+ },
+ {
+ "epoch": 0.029585798816568046,
+ "grad_norm": 0.7272042632102966,
+ "learning_rate": 7.351351351351352e-05,
+ "loss": 0.12121891975402832,
+ "num_input_tokens_seen": 117886,
+ "step": 125,
+ "train_runtime": 465.0328,
+ "train_tokens_per_second": 253.5
+ },
+ {
+ "epoch": 0.029822485207100593,
+ "grad_norm": 0.6021063923835754,
+ "learning_rate": 7.243243243243245e-05,
+ "loss": 0.08234579861164093,
+ "num_input_tokens_seen": 118806,
+ "step": 126,
+ "train_runtime": 468.1518,
+ "train_tokens_per_second": 253.777
+ },
+ {
+ "epoch": 0.030059171597633137,
+ "grad_norm": 0.5659137964248657,
+ "learning_rate": 7.135135135135136e-05,
+ "loss": 0.07469362020492554,
+ "num_input_tokens_seen": 119716,
+ "step": 127,
+ "train_runtime": 471.2559,
+ "train_tokens_per_second": 254.036
+ },
+ {
+ "epoch": 0.03029585798816568,
+ "grad_norm": 0.7604466080665588,
+ "learning_rate": 7.027027027027028e-05,
+ "loss": 0.15167035162448883,
+ "num_input_tokens_seen": 120710,
+ "step": 128,
+ "train_runtime": 474.3536,
+ "train_tokens_per_second": 254.473
+ },
+ {
+ "epoch": 0.030532544378698224,
+ "grad_norm": 0.898281455039978,
+ "learning_rate": 6.91891891891892e-05,
+ "loss": 0.18358615040779114,
+ "num_input_tokens_seen": 121668,
+ "step": 129,
+ "train_runtime": 477.4676,
+ "train_tokens_per_second": 254.819
+ },
+ {
+ "epoch": 0.03076923076923077,
+ "grad_norm": 1.046539545059204,
+ "learning_rate": 6.810810810810811e-05,
+ "loss": 0.14173445105552673,
+ "num_input_tokens_seen": 122654,
+ "step": 130,
+ "train_runtime": 480.6242,
+ "train_tokens_per_second": 255.197
+ },
+ {
+ "epoch": 0.031005917159763315,
+ "grad_norm": 0.689860463142395,
+ "learning_rate": 6.702702702702704e-05,
+ "loss": 0.1232977956533432,
+ "num_input_tokens_seen": 123610,
+ "step": 131,
+ "train_runtime": 483.7904,
+ "train_tokens_per_second": 255.503
+ },
+ {
+ "epoch": 0.031242603550295858,
+ "grad_norm": 0.6417210698127747,
+ "learning_rate": 6.594594594594596e-05,
+ "loss": 0.08111131936311722,
+ "num_input_tokens_seen": 124526,
+ "step": 132,
+ "train_runtime": 486.9241,
+ "train_tokens_per_second": 255.74
+ },
+ {
+ "epoch": 0.0314792899408284,
+ "grad_norm": 0.8019670844078064,
+ "learning_rate": 6.486486486486487e-05,
+ "loss": 0.13235916197299957,
+ "num_input_tokens_seen": 125438,
+ "step": 133,
+ "train_runtime": 490.0416,
+ "train_tokens_per_second": 255.974
+ },
+ {
+ "epoch": 0.031715976331360946,
+ "grad_norm": 0.8618372082710266,
+ "learning_rate": 6.378378378378379e-05,
+ "loss": 0.11209192872047424,
+ "num_input_tokens_seen": 126378,
+ "step": 134,
+ "train_runtime": 493.0816,
+ "train_tokens_per_second": 256.302
+ },
+ {
+ "epoch": 0.03195266272189349,
+ "grad_norm": 0.6566123962402344,
+ "learning_rate": 6.27027027027027e-05,
+ "loss": 0.08107412606477737,
+ "num_input_tokens_seen": 127280,
+ "step": 135,
+ "train_runtime": 496.1373,
+ "train_tokens_per_second": 256.542
+ },
+ {
+ "epoch": 0.03218934911242603,
+ "grad_norm": 0.671886682510376,
+ "learning_rate": 6.162162162162163e-05,
+ "loss": 0.10395485162734985,
+ "num_input_tokens_seen": 128246,
+ "step": 136,
+ "train_runtime": 499.2198,
+ "train_tokens_per_second": 256.893
+ },
+ {
+ "epoch": 0.032426035502958576,
+ "grad_norm": 0.5853029489517212,
+ "learning_rate": 6.0540540540540543e-05,
+ "loss": 0.06830425560474396,
+ "num_input_tokens_seen": 129142,
+ "step": 137,
+ "train_runtime": 502.2719,
+ "train_tokens_per_second": 257.116
+ },
+ {
+ "epoch": 0.03266272189349113,
+ "grad_norm": 0.9871621131896973,
+ "learning_rate": 5.9459459459459466e-05,
+ "loss": 0.14138084650039673,
+ "num_input_tokens_seen": 130060,
+ "step": 138,
+ "train_runtime": 505.3687,
+ "train_tokens_per_second": 257.357
+ },
+ {
+ "epoch": 0.03289940828402367,
+ "grad_norm": 0.6716732978820801,
+ "learning_rate": 5.837837837837838e-05,
+ "loss": 0.07525799423456192,
+ "num_input_tokens_seen": 130982,
+ "step": 139,
+ "train_runtime": 508.4248,
+ "train_tokens_per_second": 257.623
+ },
+ {
+ "epoch": 0.033136094674556214,
+ "grad_norm": 0.7177789807319641,
+ "learning_rate": 5.7297297297297305e-05,
+ "loss": 0.09517940878868103,
+ "num_input_tokens_seen": 131900,
+ "step": 140,
+ "train_runtime": 511.57,
+ "train_tokens_per_second": 257.834
+ },
+ {
+ "epoch": 0.03337278106508876,
+ "grad_norm": 0.703551709651947,
+ "learning_rate": 5.621621621621622e-05,
+ "loss": 0.09361469000577927,
+ "num_input_tokens_seen": 132890,
+ "step": 141,
+ "train_runtime": 514.7186,
+ "train_tokens_per_second": 258.18
+ },
+ {
+ "epoch": 0.0336094674556213,
+ "grad_norm": 0.7156357765197754,
+ "learning_rate": 5.5135135135135144e-05,
+ "loss": 0.08832181245088577,
+ "num_input_tokens_seen": 133810,
+ "step": 142,
+ "train_runtime": 517.7644,
+ "train_tokens_per_second": 258.438
+ },
+ {
+ "epoch": 0.033846153846153845,
+ "grad_norm": 0.7757886648178101,
+ "learning_rate": 5.405405405405406e-05,
+ "loss": 0.12478049844503403,
+ "num_input_tokens_seen": 134758,
+ "step": 143,
+ "train_runtime": 520.8215,
+ "train_tokens_per_second": 258.741
+ },
+ {
+ "epoch": 0.03408284023668639,
+ "grad_norm": 0.6047410368919373,
+ "learning_rate": 5.2972972972972976e-05,
+ "loss": 0.07645335048437119,
+ "num_input_tokens_seen": 135724,
+ "step": 144,
+ "train_runtime": 523.9787,
+ "train_tokens_per_second": 259.026
+ },
+ {
+ "epoch": 0.03431952662721893,
+ "grad_norm": 0.7750188708305359,
+ "learning_rate": 5.18918918918919e-05,
+ "loss": 0.10904290527105331,
+ "num_input_tokens_seen": 136662,
+ "step": 145,
+ "train_runtime": 527.0633,
+ "train_tokens_per_second": 259.29
+ },
+ {
+ "epoch": 0.034556213017751476,
+ "grad_norm": 0.6400149464607239,
+ "learning_rate": 5.0810810810810815e-05,
+ "loss": 0.07525563985109329,
+ "num_input_tokens_seen": 137628,
+ "step": 146,
+ "train_runtime": 530.1299,
+ "train_tokens_per_second": 259.612
+ },
+ {
+ "epoch": 0.03479289940828403,
+ "grad_norm": 0.6681646108627319,
+ "learning_rate": 4.972972972972974e-05,
+ "loss": 0.08536849915981293,
+ "num_input_tokens_seen": 138610,
+ "step": 147,
+ "train_runtime": 533.159,
+ "train_tokens_per_second": 259.979
+ },
+ {
+ "epoch": 0.03502958579881657,
+ "grad_norm": 0.6158885359764099,
+ "learning_rate": 4.8648648648648654e-05,
+ "loss": 0.0753958523273468,
+ "num_input_tokens_seen": 139486,
+ "step": 148,
+ "train_runtime": 536.2632,
+ "train_tokens_per_second": 260.107
+ },
+ {
+ "epoch": 0.035266272189349114,
+ "grad_norm": 0.665565550327301,
+ "learning_rate": 4.756756756756757e-05,
+ "loss": 0.08198000490665436,
+ "num_input_tokens_seen": 140440,
+ "step": 149,
+ "train_runtime": 539.3165,
+ "train_tokens_per_second": 260.404
+ },
+ {
+ "epoch": 0.03550295857988166,
+ "grad_norm": 0.6818353533744812,
+ "learning_rate": 4.648648648648649e-05,
+ "loss": 0.0710655227303505,
+ "num_input_tokens_seen": 141344,
+ "step": 150,
+ "train_runtime": 542.3824,
+ "train_tokens_per_second": 260.598
+ },
+ {
+ "epoch": 0.0357396449704142,
+ "grad_norm": 0.5292275547981262,
+ "learning_rate": 4.540540540540541e-05,
+ "loss": 0.04743894189596176,
+ "num_input_tokens_seen": 142242,
+ "step": 151,
+ "train_runtime": 549.1326,
+ "train_tokens_per_second": 259.03
+ },
+ {
+ "epoch": 0.035976331360946745,
+ "grad_norm": 0.9436084032058716,
+ "learning_rate": 4.432432432432433e-05,
+ "loss": 0.10009428858757019,
+ "num_input_tokens_seen": 143104,
+ "step": 152,
+ "train_runtime": 552.245,
+ "train_tokens_per_second": 259.131
+ },
+ {
+ "epoch": 0.03621301775147929,
+ "grad_norm": 0.8146379590034485,
+ "learning_rate": 4.324324324324325e-05,
+ "loss": 0.08397847414016724,
+ "num_input_tokens_seen": 144036,
+ "step": 153,
+ "train_runtime": 555.3194,
+ "train_tokens_per_second": 259.375
+ },
+ {
+ "epoch": 0.03644970414201183,
+ "grad_norm": 0.6370963454246521,
+ "learning_rate": 4.2162162162162164e-05,
+ "loss": 0.0647936463356018,
+ "num_input_tokens_seen": 144978,
+ "step": 154,
+ "train_runtime": 558.5158,
+ "train_tokens_per_second": 259.577
+ },
+ {
+ "epoch": 0.036686390532544376,
+ "grad_norm": 0.7876921892166138,
+ "learning_rate": 4.108108108108109e-05,
+ "loss": 0.10044769942760468,
+ "num_input_tokens_seen": 145932,
+ "step": 155,
+ "train_runtime": 561.6626,
+ "train_tokens_per_second": 259.821
+ },
+ {
+ "epoch": 0.036923076923076927,
+ "grad_norm": 0.5711075067520142,
+ "learning_rate": 4e-05,
+ "loss": 0.0628824308514595,
+ "num_input_tokens_seen": 146822,
+ "step": 156,
+ "train_runtime": 564.7686,
+ "train_tokens_per_second": 259.968
+ },
+ {
+ "epoch": 0.03715976331360947,
+ "grad_norm": 0.6890460848808289,
+ "learning_rate": 3.8918918918918926e-05,
+ "loss": 0.0714673101902008,
+ "num_input_tokens_seen": 147746,
+ "step": 157,
+ "train_runtime": 567.7959,
+ "train_tokens_per_second": 260.21
+ },
+ {
+ "epoch": 0.037396449704142014,
+ "grad_norm": 0.6936295032501221,
+ "learning_rate": 3.783783783783784e-05,
+ "loss": 0.05670904368162155,
+ "num_input_tokens_seen": 148648,
+ "step": 158,
+ "train_runtime": 570.8255,
+ "train_tokens_per_second": 260.409
+ },
+ {
+ "epoch": 0.03763313609467456,
+ "grad_norm": 0.6268077492713928,
+ "learning_rate": 3.675675675675676e-05,
+ "loss": 0.04924366623163223,
+ "num_input_tokens_seen": 149622,
+ "step": 159,
+ "train_runtime": 573.9724,
+ "train_tokens_per_second": 260.678
+ },
+ {
+ "epoch": 0.0378698224852071,
+ "grad_norm": 0.7037569284439087,
+ "learning_rate": 3.567567567567568e-05,
+ "loss": 0.06151504069566727,
+ "num_input_tokens_seen": 150586,
+ "step": 160,
+ "train_runtime": 577.103,
+ "train_tokens_per_second": 260.934
+ },
+ {
+ "epoch": 0.038106508875739645,
+ "grad_norm": 0.6471953392028809,
+ "learning_rate": 3.45945945945946e-05,
+ "loss": 0.040921665728092194,
+ "num_input_tokens_seen": 151442,
+ "step": 161,
+ "train_runtime": 580.1219,
+ "train_tokens_per_second": 261.052
+ },
+ {
+ "epoch": 0.03834319526627219,
+ "grad_norm": 0.7046090960502625,
+ "learning_rate": 3.351351351351352e-05,
+ "loss": 0.05204992741346359,
+ "num_input_tokens_seen": 152350,
+ "step": 162,
+ "train_runtime": 583.1537,
+ "train_tokens_per_second": 261.252
+ },
+ {
+ "epoch": 0.03857988165680473,
+ "grad_norm": 1.0495421886444092,
+ "learning_rate": 3.2432432432432436e-05,
+ "loss": 0.1057022288441658,
+ "num_input_tokens_seen": 153278,
+ "step": 163,
+ "train_runtime": 586.3134,
+ "train_tokens_per_second": 261.427
+ },
+ {
+ "epoch": 0.038816568047337276,
+ "grad_norm": 0.6432052850723267,
+ "learning_rate": 3.135135135135135e-05,
+ "loss": 0.05509451404213905,
+ "num_input_tokens_seen": 154178,
+ "step": 164,
+ "train_runtime": 589.3544,
+ "train_tokens_per_second": 261.605
+ },
+ {
+ "epoch": 0.03905325443786982,
+ "grad_norm": 0.6547747254371643,
+ "learning_rate": 3.0270270270270272e-05,
+ "loss": 0.04602127522230148,
+ "num_input_tokens_seen": 155080,
+ "step": 165,
+ "train_runtime": 592.3642,
+ "train_tokens_per_second": 261.798
+ },
+ {
+ "epoch": 0.03928994082840237,
+ "grad_norm": 0.6700029969215393,
+ "learning_rate": 2.918918918918919e-05,
+ "loss": 0.05721529945731163,
+ "num_input_tokens_seen": 156140,
+ "step": 166,
+ "train_runtime": 595.4897,
+ "train_tokens_per_second": 262.204
+ },
+ {
+ "epoch": 0.039526627218934914,
+ "grad_norm": 0.9145491719245911,
+ "learning_rate": 2.810810810810811e-05,
+ "loss": 0.06747015565633774,
+ "num_input_tokens_seen": 157106,
+ "step": 167,
+ "train_runtime": 598.5891,
+ "train_tokens_per_second": 262.461
+ },
+ {
+ "epoch": 0.03976331360946746,
+ "grad_norm": 0.8798087239265442,
+ "learning_rate": 2.702702702702703e-05,
+ "loss": 0.08232270181179047,
+ "num_input_tokens_seen": 158056,
+ "step": 168,
+ "train_runtime": 601.7315,
+ "train_tokens_per_second": 262.669
+ },
+ {
+ "epoch": 0.04,
+ "grad_norm": 1.0618623495101929,
+ "learning_rate": 2.594594594594595e-05,
+ "loss": 0.09569647908210754,
+ "num_input_tokens_seen": 159060,
+ "step": 169,
+ "train_runtime": 604.8838,
+ "train_tokens_per_second": 262.96
+ },
+ {
+ "epoch": 0.040236686390532544,
+ "grad_norm": 0.6945156455039978,
+ "learning_rate": 2.486486486486487e-05,
+ "loss": 0.0597086027264595,
+ "num_input_tokens_seen": 160054,
+ "step": 170,
+ "train_runtime": 608.004,
+ "train_tokens_per_second": 263.245
+ },
+ {
+ "epoch": 0.04047337278106509,
+ "grad_norm": 0.6867018342018127,
+ "learning_rate": 2.3783783783783785e-05,
+ "loss": 0.053483735769987106,
+ "num_input_tokens_seen": 160992,
+ "step": 171,
+ "train_runtime": 611.0992,
+ "train_tokens_per_second": 263.447
+ },
+ {
+ "epoch": 0.04071005917159763,
+ "grad_norm": 0.9504451751708984,
+ "learning_rate": 2.2702702702702705e-05,
+ "loss": 0.07532191276550293,
+ "num_input_tokens_seen": 161910,
+ "step": 172,
+ "train_runtime": 614.2523,
+ "train_tokens_per_second": 263.589
+ },
+ {
+ "epoch": 0.040946745562130175,
+ "grad_norm": 0.6639146208763123,
+ "learning_rate": 2.1621621621621624e-05,
+ "loss": 0.04415765404701233,
+ "num_input_tokens_seen": 162810,
+ "step": 173,
+ "train_runtime": 617.2712,
+ "train_tokens_per_second": 263.758
+ },
+ {
+ "epoch": 0.04118343195266272,
+ "grad_norm": 0.5948973894119263,
+ "learning_rate": 2.0540540540540544e-05,
+ "loss": 0.04589955136179924,
+ "num_input_tokens_seen": 163792,
+ "step": 174,
+ "train_runtime": 620.3367,
+ "train_tokens_per_second": 264.037
+ },
+ {
+ "epoch": 0.04142011834319527,
+ "grad_norm": 0.8319454193115234,
+ "learning_rate": 1.9459459459459463e-05,
+ "loss": 0.07309269905090332,
+ "num_input_tokens_seen": 164770,
+ "step": 175,
+ "train_runtime": 623.4097,
+ "train_tokens_per_second": 264.305
+ },
+ {
+ "epoch": 0.04165680473372781,
+ "grad_norm": 0.591090202331543,
+ "learning_rate": 1.837837837837838e-05,
+ "loss": 0.04063474386930466,
+ "num_input_tokens_seen": 165708,
+ "step": 176,
+ "train_runtime": 626.513,
+ "train_tokens_per_second": 264.493
+ },
+ {
+ "epoch": 0.04189349112426036,
+ "grad_norm": 0.5455971360206604,
+ "learning_rate": 1.72972972972973e-05,
+ "loss": 0.0511426106095314,
+ "num_input_tokens_seen": 166654,
+ "step": 177,
+ "train_runtime": 629.5606,
+ "train_tokens_per_second": 264.715
+ },
+ {
+ "epoch": 0.0421301775147929,
+ "grad_norm": 0.7040805816650391,
+ "learning_rate": 1.6216216216216218e-05,
+ "loss": 0.06684890389442444,
+ "num_input_tokens_seen": 167586,
+ "step": 178,
+ "train_runtime": 632.5911,
+ "train_tokens_per_second": 264.92
+ },
+ {
+ "epoch": 0.042366863905325444,
+ "grad_norm": 0.9840870499610901,
+ "learning_rate": 1.5135135135135136e-05,
+ "loss": 0.08544427156448364,
+ "num_input_tokens_seen": 168558,
+ "step": 179,
+ "train_runtime": 635.6343,
+ "train_tokens_per_second": 265.181
+ },
+ {
+ "epoch": 0.04260355029585799,
+ "grad_norm": 0.763282835483551,
+ "learning_rate": 1.4054054054054055e-05,
+ "loss": 0.05402141064405441,
+ "num_input_tokens_seen": 169594,
+ "step": 180,
+ "train_runtime": 638.8133,
+ "train_tokens_per_second": 265.483
+ }
+ ],
+ "logging_steps": 1,
+ "max_steps": 190,
+ "num_input_tokens_seen": 169594,
+ "num_train_epochs": 1,
+ "save_steps": 30,
+ "stateful_callbacks": {
+ "TrainerControl": {
+ "args": {
+ "should_epoch_stop": false,
+ "should_evaluate": false,
+ "should_log": false,
+ "should_save": true,
+ "should_training_stop": false
+ },
+ "attributes": {}
+ }
+ },
+ "total_flos": 2444956052661120.0,
+ "train_batch_size": 2,
+ "trial_name": null,
+ "trial_params": null
+}
diff --git a/checkpoint-180/training_args.bin b/checkpoint-180/training_args.bin
new file mode 100644
index 0000000000000000000000000000000000000000..cefffe2bb567bdf18003e5cc5aaf746472c2981f
--- /dev/null
+++ b/checkpoint-180/training_args.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:34c4ca6d50f04301fb38588516a91421e14a8ad1ffddf12afdfd9ba35cd4b64e
+size 5777
diff --git a/checkpoint-190/README.md b/checkpoint-190/README.md
new file mode 100644
index 0000000000000000000000000000000000000000..41470ec6557bbc37155893f70bd59986401b9148
--- /dev/null
+++ b/checkpoint-190/README.md
@@ -0,0 +1,210 @@
+---
+base_model: unsloth/gemma-4-e2b-it-unsloth-bnb-4bit
+library_name: peft
+pipeline_tag: text-generation
+tags:
+- base_model:adapter:unsloth/gemma-4-e2b-it-unsloth-bnb-4bit
+- lora
+- sft
+- transformers
+- trl
+- unsloth
+---
+
+# Model Card for Model ID
+
+
+
+
+
+## Model Details
+
+### Model Description
+
+
+
+
+
+- **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]
+
+
+
+- **Repository:** [More Information Needed]
+- **Paper [optional]:** [More Information Needed]
+- **Demo [optional]:** [More Information Needed]
+
+## Uses
+
+
+
+### Direct Use
+
+
+
+[More Information Needed]
+
+### Downstream Use [optional]
+
+
+
+[More Information Needed]
+
+### Out-of-Scope Use
+
+
+
+[More Information Needed]
+
+## Bias, Risks, and Limitations
+
+
+
+[More Information Needed]
+
+### Recommendations
+
+
+
+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
+
+
+
+[More Information Needed]
+
+### Training Procedure
+
+
+
+#### Preprocessing [optional]
+
+[More Information Needed]
+
+
+#### Training Hyperparameters
+
+- **Training regime:** [More Information Needed]
+
+#### Speeds, Sizes, Times [optional]
+
+
+
+[More Information Needed]
+
+## Evaluation
+
+
+
+### Testing Data, Factors & Metrics
+
+#### Testing Data
+
+
+
+[More Information Needed]
+
+#### Factors
+
+
+
+[More Information Needed]
+
+#### Metrics
+
+
+
+[More Information Needed]
+
+### Results
+
+[More Information Needed]
+
+#### Summary
+
+
+
+## Model Examination [optional]
+
+
+
+[More Information Needed]
+
+## Environmental Impact
+
+
+
+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]
+
+
+
+**BibTeX:**
+
+[More Information Needed]
+
+**APA:**
+
+[More Information Needed]
+
+## Glossary [optional]
+
+
+
+[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
\ No newline at end of file
diff --git a/checkpoint-190/adapter_config.json b/checkpoint-190/adapter_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..915726be1cd67eb165724a8390b51a57fe0dd6df
--- /dev/null
+++ b/checkpoint-190/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": "unsloth/gemma-4-e2b-it-unsloth-bnb-4bit",
+ "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": [
+ "q_proj",
+ "gate_proj",
+ "v_proj",
+ "k_proj",
+ "down_proj",
+ "o_proj",
+ "up_proj"
+ ],
+ "target_parameters": null,
+ "task_type": "CAUSAL_LM",
+ "trainable_token_indices": null,
+ "use_dora": false,
+ "use_qalora": false,
+ "use_rslora": false
+}
\ No newline at end of file
diff --git a/checkpoint-190/adapter_model.safetensors b/checkpoint-190/adapter_model.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..2c87a0ef165f037ee646d9506ba63a7bcc4aa079
--- /dev/null
+++ b/checkpoint-190/adapter_model.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:dae7c101be06c818341a1b3354a5715c93ad40fc4c4700a81f0c71c0e699b82b
+size 124277728
diff --git a/checkpoint-190/chat_template.jinja b/checkpoint-190/chat_template.jinja
new file mode 100644
index 0000000000000000000000000000000000000000..dc032e46e28e0d325b18d26da1279cdda3afa8d7
--- /dev/null
+++ b/checkpoint-190/chat_template.jinja
@@ -0,0 +1,351 @@
+{%- macro format_parameters(properties, required, filter_keys=false) -%}
+ {%- 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 not filter_keys or 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([]), filter_keys=true) -}}
+ }
+ {%- 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('') -%}
+ {%- 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 -%}
+ {{- '' -}}
+{%- 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'] -%}
+ {%- if messages[0]['content'] is string -%}
+ {{- messages[0]['content'] | trim -}}
+ {%- elif messages[0]['content'] is sequence -%}
+ {%- for item in messages[0]['content'] -%}
+ {{- item['text'] | trim + ' '-}}
+ {%- endfor -%}
+ {%- endif -%}
+ {%- set loop_messages = messages[1:] -%}
+ {%- endif -%}
+ {%- if tools -%}
+ {%- for tool in tools %}
+ {{- '<|tool>' -}}
+ {{- format_function_declaration(tool) | trim -}}
+ {{- '' -}}
+ {%- endfor %}
+ {%- set ns.prev_message_type = 'tool' -%}
+ {%- endif -%}
+ {{- '\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' -}}
+ {%- 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 -%}
+ {{- '}' -}}
+ {%- 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 -%}
+
+ {%- set captured_content -%}
+ {%- 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 -%}
+ {%- endset -%}
+
+ {{- captured_content -}}
+ {%- set has_content = captured_content | trim | length > 0 -%}
+
+ {%- if ns.prev_message_type == 'tool_call' and not ns_tr_out.flag -%}
+ {{- '<|tool_response>' -}}
+ {%- elif not (ns_tr_out.flag and not has_content) -%}
+ {{- '\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 -%}
\ No newline at end of file
diff --git a/checkpoint-190/optimizer.pt b/checkpoint-190/optimizer.pt
new file mode 100644
index 0000000000000000000000000000000000000000..6e0dff6925573d76e7fc90a716316ed2ac69af96
--- /dev/null
+++ b/checkpoint-190/optimizer.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b26ffa09ce09033bd748c71a5bd466483b607675e03fe146c587da6439440b9d
+size 52047117
diff --git a/checkpoint-190/processor_config.json b/checkpoint-190/processor_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..bfaa5e360ad17c2761b7ee3b05abf03e381e8aa1
--- /dev/null
+++ b/checkpoint-190/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"
+ }
+}
diff --git a/checkpoint-190/rng_state.pth b/checkpoint-190/rng_state.pth
new file mode 100644
index 0000000000000000000000000000000000000000..3ef66339b9befa098183fd5d69faed6838e526b0
--- /dev/null
+++ b/checkpoint-190/rng_state.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f1d565802a8e26c4e8a31328752b7a7fdc186d9401aa008e65697d0ad8c22e33
+size 14645
diff --git a/checkpoint-190/scaler.pt b/checkpoint-190/scaler.pt
new file mode 100644
index 0000000000000000000000000000000000000000..8f1374651b0fad13c4f866d696a73d7b6b357f25
--- /dev/null
+++ b/checkpoint-190/scaler.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e582bf3c00963bd3fcaa4b162e161d1ea274c715ce05768df1a86c80c59a424e
+size 1383
diff --git a/checkpoint-190/scheduler.pt b/checkpoint-190/scheduler.pt
new file mode 100644
index 0000000000000000000000000000000000000000..0409ba8b86a1aaa959f73dee2211a8e734a66469
--- /dev/null
+++ b/checkpoint-190/scheduler.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:30ec32e5e6be2d8912b2798a0a556d384a766d11fbf712d461217391922e67ba
+size 1465
diff --git a/checkpoint-190/tokenizer.json b/checkpoint-190/tokenizer.json
new file mode 100644
index 0000000000000000000000000000000000000000..1ff9f3e3439a939b971f9919e821bf87e835a503
--- /dev/null
+++ b/checkpoint-190/tokenizer.json
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:cc8d3a0ce36466ccc1278bf987df5f71db1719b9ca6b4118264f45cb627bfe0f
+size 32169626
diff --git a/checkpoint-190/tokenizer_config.json b/checkpoint-190/tokenizer_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..24b1522bb32b0eb05b631b1a9f112751b8cb0fab
--- /dev/null
+++ b/checkpoint-190/tokenizer_config.json
@@ -0,0 +1,289 @@
+{
+ "audio_token": "<|audio|>",
+ "backend": "tokenizers",
+ "boa_token": "<|audio>",
+ "boi_token": "<|image>",
+ "bos_token": "",
+ "eoa_token": "",
+ "eoc_token": "",
+ "eoi_token": "",
+ "eos_token": "",
+ "eot_token": "",
+ "escape_token": "<|\"|>",
+ "etc_token": "",
+ "etd_token": "",
+ "etr_token": "",
+ "extra_special_tokens": [
+ "<|video|>"
+ ],
+ "image_token": "<|image|>",
+ "is_local": false,
+ "mask_token": "",
+ "model_max_length": 131072,
+ "model_specific_special_tokens": {
+ "audio_token": "<|audio|>",
+ "boa_token": "<|audio>",
+ "boi_token": "<|image>",
+ "eoa_token": "",
+ "eoc_token": "",
+ "eoi_token": "",
+ "eot_token": "",
+ "escape_token": "<|\"|>",
+ "etc_token": "",
+ "etd_token": "",
+ "etr_token": "",
+ "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": "",
+ "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\\w+)(?P\\{.*\\})"
+ },
+ "type": {
+ "const": "function"
+ }
+ },
+ "type": "object"
+ },
+ "type": "array",
+ "x-regex-iterator": "<\\|tool_call>(.*?)"
+ }
+ },
+ "type": "object",
+ "x-regex": "(\\<\\|channel\\>thought\\n(?P.*?)\\)?(?P\\<\\|tool_call\\>.*\\)?(?P(?:(?!\\)(?!\\<\\|tool_response\\>).)+)?(?:\\|\\<\\|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": "",
+ "added_tokens_decoder": {
+ "0": {
+ "content": "",
+ "single_word": false,
+ "lstrip": false,
+ "rstrip": false,
+ "normalized": false,
+ "special": true
+ },
+ "1": {
+ "content": "",
+ "single_word": false,
+ "lstrip": false,
+ "rstrip": false,
+ "normalized": false,
+ "special": true
+ },
+ "2": {
+ "content": "",
+ "single_word": false,
+ "lstrip": false,
+ "rstrip": false,
+ "normalized": false,
+ "special": true
+ },
+ "3": {
+ "content": "",
+ "single_word": false,
+ "lstrip": false,
+ "rstrip": false,
+ "normalized": false,
+ "special": true
+ },
+ "4": {
+ "content": "",
+ "single_word": false,
+ "lstrip": false,
+ "rstrip": false,
+ "normalized": false,
+ "special": true
+ },
+ "46": {
+ "content": "<|tool>",
+ "single_word": false,
+ "lstrip": false,
+ "rstrip": false,
+ "normalized": false,
+ "special": true
+ },
+ "47": {
+ "content": "",
+ "single_word": false,
+ "lstrip": false,
+ "rstrip": false,
+ "normalized": false,
+ "special": true
+ },
+ "48": {
+ "content": "<|tool_call>",
+ "single_word": false,
+ "lstrip": false,
+ "rstrip": false,
+ "normalized": false,
+ "special": true
+ },
+ "49": {
+ "content": "",
+ "single_word": false,
+ "lstrip": false,
+ "rstrip": false,
+ "normalized": false,
+ "special": true
+ },
+ "50": {
+ "content": "<|tool_response>",
+ "single_word": false,
+ "lstrip": false,
+ "rstrip": false,
+ "normalized": false,
+ "special": true
+ },
+ "51": {
+ "content": "",
+ "single_word": false,
+ "lstrip": false,
+ "rstrip": false,
+ "normalized": false,
+ "special": true
+ },
+ "52": {
+ "content": "<|\"|>",
+ "single_word": false,
+ "lstrip": false,
+ "rstrip": false,
+ "normalized": false,
+ "special": true
+ },
+ "98": {
+ "content": "<|think|>",
+ "single_word": false,
+ "lstrip": false,
+ "rstrip": false,
+ "normalized": false,
+ "special": true
+ },
+ "100": {
+ "content": "<|channel>",
+ "single_word": false,
+ "lstrip": false,
+ "rstrip": false,
+ "normalized": false,
+ "special": true
+ },
+ "101": {
+ "content": "",
+ "single_word": false,
+ "lstrip": false,
+ "rstrip": false,
+ "normalized": false,
+ "special": true
+ },
+ "105": {
+ "content": "<|turn>",
+ "single_word": false,
+ "lstrip": false,
+ "rstrip": false,
+ "normalized": false,
+ "special": true
+ },
+ "106": {
+ "content": "",
+ "single_word": false,
+ "lstrip": false,
+ "rstrip": false,
+ "normalized": false,
+ "special": true
+ },
+ "255999": {
+ "content": "<|image>",
+ "single_word": false,
+ "lstrip": false,
+ "rstrip": false,
+ "normalized": false,
+ "special": true
+ },
+ "256000": {
+ "content": "<|audio>",
+ "single_word": false,
+ "lstrip": false,
+ "rstrip": false,
+ "normalized": false,
+ "special": true
+ },
+ "258880": {
+ "content": "<|image|>",
+ "single_word": false,
+ "lstrip": false,
+ "rstrip": false,
+ "normalized": false,
+ "special": true
+ },
+ "258881": {
+ "content": "<|audio|>",
+ "single_word": false,
+ "lstrip": false,
+ "rstrip": false,
+ "normalized": false,
+ "special": true
+ },
+ "258882": {
+ "content": "",
+ "single_word": false,
+ "lstrip": false,
+ "rstrip": false,
+ "normalized": false,
+ "special": true
+ },
+ "258883": {
+ "content": "",
+ "single_word": false,
+ "lstrip": false,
+ "rstrip": false,
+ "normalized": false,
+ "special": true
+ },
+ "258884": {
+ "content": "<|video|>",
+ "single_word": false,
+ "lstrip": false,
+ "rstrip": false,
+ "normalized": false,
+ "special": true
+ }
+ }
+}
diff --git a/checkpoint-190/trainer_state.json b/checkpoint-190/trainer_state.json
new file mode 100644
index 0000000000000000000000000000000000000000..ad0a1ca66a30d7c619cfb869f44a57f7a745fd27
--- /dev/null
+++ b/checkpoint-190/trainer_state.json
@@ -0,0 +1,1934 @@
+{
+ "best_global_step": null,
+ "best_metric": null,
+ "best_model_checkpoint": null,
+ "epoch": 0.04497041420118343,
+ "eval_steps": 500,
+ "global_step": 190,
+ "is_hyper_param_search": false,
+ "is_local_process_zero": true,
+ "is_world_process_zero": true,
+ "log_history": [
+ {
+ "epoch": 0.00023668639053254438,
+ "grad_norm": NaN,
+ "learning_rate": 0.0,
+ "loss": 4.347962379455566,
+ "num_input_tokens_seen": 842,
+ "step": 1,
+ "train_runtime": 63.0289,
+ "train_tokens_per_second": 13.359
+ },
+ {
+ "epoch": 0.00047337278106508875,
+ "grad_norm": NaN,
+ "learning_rate": 0.0,
+ "loss": 4.634688854217529,
+ "num_input_tokens_seen": 1820,
+ "step": 2,
+ "train_runtime": 67.7033,
+ "train_tokens_per_second": 26.882
+ },
+ {
+ "epoch": 0.0007100591715976331,
+ "grad_norm": 13.944212913513184,
+ "learning_rate": 0.0,
+ "loss": 4.42241096496582,
+ "num_input_tokens_seen": 2754,
+ "step": 3,
+ "train_runtime": 72.324,
+ "train_tokens_per_second": 38.079
+ },
+ {
+ "epoch": 0.0009467455621301775,
+ "grad_norm": 13.445413589477539,
+ "learning_rate": 4e-05,
+ "loss": 4.682804107666016,
+ "num_input_tokens_seen": 3700,
+ "step": 4,
+ "train_runtime": 75.3925,
+ "train_tokens_per_second": 49.077
+ },
+ {
+ "epoch": 0.001183431952662722,
+ "grad_norm": 10.576727867126465,
+ "learning_rate": 8e-05,
+ "loss": 4.166159629821777,
+ "num_input_tokens_seen": 4714,
+ "step": 5,
+ "train_runtime": 78.474,
+ "train_tokens_per_second": 60.071
+ },
+ {
+ "epoch": 0.0014201183431952662,
+ "grad_norm": 8.077850341796875,
+ "learning_rate": 0.00012,
+ "loss": 3.2006261348724365,
+ "num_input_tokens_seen": 5674,
+ "step": 6,
+ "train_runtime": 81.6139,
+ "train_tokens_per_second": 69.522
+ },
+ {
+ "epoch": 0.0016568047337278107,
+ "grad_norm": 6.60789680480957,
+ "learning_rate": 0.00016,
+ "loss": 2.7067415714263916,
+ "num_input_tokens_seen": 6646,
+ "step": 7,
+ "train_runtime": 84.786,
+ "train_tokens_per_second": 78.386
+ },
+ {
+ "epoch": 0.001893491124260355,
+ "grad_norm": 9.175209045410156,
+ "learning_rate": 0.0002,
+ "loss": 2.367570638656616,
+ "num_input_tokens_seen": 7518,
+ "step": 8,
+ "train_runtime": 87.8178,
+ "train_tokens_per_second": 85.609
+ },
+ {
+ "epoch": 0.0021301775147928993,
+ "grad_norm": 3.438948631286621,
+ "learning_rate": 0.00019891891891891895,
+ "loss": 2.1013119220733643,
+ "num_input_tokens_seen": 8470,
+ "step": 9,
+ "train_runtime": 90.9458,
+ "train_tokens_per_second": 93.132
+ },
+ {
+ "epoch": 0.002366863905325444,
+ "grad_norm": 3.459472894668579,
+ "learning_rate": 0.00019783783783783784,
+ "loss": 1.8991940021514893,
+ "num_input_tokens_seen": 9458,
+ "step": 10,
+ "train_runtime": 94.0577,
+ "train_tokens_per_second": 100.555
+ },
+ {
+ "epoch": 0.0026035502958579883,
+ "grad_norm": 2.169750928878784,
+ "learning_rate": 0.00019675675675675678,
+ "loss": 1.7395055294036865,
+ "num_input_tokens_seen": 10456,
+ "step": 11,
+ "train_runtime": 97.2068,
+ "train_tokens_per_second": 107.565
+ },
+ {
+ "epoch": 0.0028402366863905324,
+ "grad_norm": 1.8638927936553955,
+ "learning_rate": 0.00019567567567567567,
+ "loss": 1.5622950792312622,
+ "num_input_tokens_seen": 11410,
+ "step": 12,
+ "train_runtime": 100.2911,
+ "train_tokens_per_second": 113.769
+ },
+ {
+ "epoch": 0.003076923076923077,
+ "grad_norm": 1.4741116762161255,
+ "learning_rate": 0.00019459459459459462,
+ "loss": 1.392699956893921,
+ "num_input_tokens_seen": 12316,
+ "step": 13,
+ "train_runtime": 103.3664,
+ "train_tokens_per_second": 119.149
+ },
+ {
+ "epoch": 0.0033136094674556214,
+ "grad_norm": 1.3238699436187744,
+ "learning_rate": 0.00019351351351351353,
+ "loss": 1.2535074949264526,
+ "num_input_tokens_seen": 13256,
+ "step": 14,
+ "train_runtime": 106.4031,
+ "train_tokens_per_second": 124.583
+ },
+ {
+ "epoch": 0.0035502958579881655,
+ "grad_norm": 1.276484489440918,
+ "learning_rate": 0.00019243243243243245,
+ "loss": 1.135048508644104,
+ "num_input_tokens_seen": 14228,
+ "step": 15,
+ "train_runtime": 109.4909,
+ "train_tokens_per_second": 129.947
+ },
+ {
+ "epoch": 0.00378698224852071,
+ "grad_norm": 1.1560865640640259,
+ "learning_rate": 0.00019135135135135137,
+ "loss": 1.0317097902297974,
+ "num_input_tokens_seen": 15192,
+ "step": 16,
+ "train_runtime": 112.5741,
+ "train_tokens_per_second": 134.951
+ },
+ {
+ "epoch": 0.004023668639053254,
+ "grad_norm": 1.1331353187561035,
+ "learning_rate": 0.00019027027027027028,
+ "loss": 0.9967219233512878,
+ "num_input_tokens_seen": 16130,
+ "step": 17,
+ "train_runtime": 115.6495,
+ "train_tokens_per_second": 139.473
+ },
+ {
+ "epoch": 0.004260355029585799,
+ "grad_norm": 1.1567702293395996,
+ "learning_rate": 0.0001891891891891892,
+ "loss": 0.905368447303772,
+ "num_input_tokens_seen": 17078,
+ "step": 18,
+ "train_runtime": 118.6796,
+ "train_tokens_per_second": 143.9
+ },
+ {
+ "epoch": 0.004497041420118343,
+ "grad_norm": 0.9157600998878479,
+ "learning_rate": 0.00018810810810810814,
+ "loss": 0.8075690865516663,
+ "num_input_tokens_seen": 18062,
+ "step": 19,
+ "train_runtime": 121.8602,
+ "train_tokens_per_second": 148.219
+ },
+ {
+ "epoch": 0.004733727810650888,
+ "grad_norm": 0.8306605815887451,
+ "learning_rate": 0.00018702702702702703,
+ "loss": 0.8007175922393799,
+ "num_input_tokens_seen": 19078,
+ "step": 20,
+ "train_runtime": 125.0019,
+ "train_tokens_per_second": 152.622
+ },
+ {
+ "epoch": 0.004970414201183432,
+ "grad_norm": 0.690018355846405,
+ "learning_rate": 0.00018594594594594597,
+ "loss": 0.7810791730880737,
+ "num_input_tokens_seen": 20068,
+ "step": 21,
+ "train_runtime": 128.0932,
+ "train_tokens_per_second": 156.667
+ },
+ {
+ "epoch": 0.005207100591715977,
+ "grad_norm": 1.4768867492675781,
+ "learning_rate": 0.00018486486486486486,
+ "loss": 0.7172120809555054,
+ "num_input_tokens_seen": 20972,
+ "step": 22,
+ "train_runtime": 131.0901,
+ "train_tokens_per_second": 159.982
+ },
+ {
+ "epoch": 0.00544378698224852,
+ "grad_norm": 0.7999809980392456,
+ "learning_rate": 0.0001837837837837838,
+ "loss": 0.6333726048469543,
+ "num_input_tokens_seen": 21900,
+ "step": 23,
+ "train_runtime": 134.212,
+ "train_tokens_per_second": 163.175
+ },
+ {
+ "epoch": 0.005680473372781065,
+ "grad_norm": 0.9285996556282043,
+ "learning_rate": 0.00018270270270270272,
+ "loss": 0.6750729084014893,
+ "num_input_tokens_seen": 22770,
+ "step": 24,
+ "train_runtime": 137.3193,
+ "train_tokens_per_second": 165.818
+ },
+ {
+ "epoch": 0.005917159763313609,
+ "grad_norm": 0.7733820080757141,
+ "learning_rate": 0.00018162162162162164,
+ "loss": 0.672032356262207,
+ "num_input_tokens_seen": 23746,
+ "step": 25,
+ "train_runtime": 140.3175,
+ "train_tokens_per_second": 169.23
+ },
+ {
+ "epoch": 0.006153846153846154,
+ "grad_norm": 0.73944091796875,
+ "learning_rate": 0.00018054054054054055,
+ "loss": 0.6716101169586182,
+ "num_input_tokens_seen": 24716,
+ "step": 26,
+ "train_runtime": 143.3683,
+ "train_tokens_per_second": 172.395
+ },
+ {
+ "epoch": 0.006390532544378698,
+ "grad_norm": 0.7659372091293335,
+ "learning_rate": 0.00017945945945945947,
+ "loss": 0.6067175269126892,
+ "num_input_tokens_seen": 25724,
+ "step": 27,
+ "train_runtime": 146.5377,
+ "train_tokens_per_second": 175.545
+ },
+ {
+ "epoch": 0.006627218934911243,
+ "grad_norm": 0.5428617000579834,
+ "learning_rate": 0.00017837837837837839,
+ "loss": 0.6083395481109619,
+ "num_input_tokens_seen": 26712,
+ "step": 28,
+ "train_runtime": 149.6429,
+ "train_tokens_per_second": 178.505
+ },
+ {
+ "epoch": 0.006863905325443787,
+ "grad_norm": 0.5915114283561707,
+ "learning_rate": 0.00017729729729729733,
+ "loss": 0.6463742256164551,
+ "num_input_tokens_seen": 27704,
+ "step": 29,
+ "train_runtime": 152.7501,
+ "train_tokens_per_second": 181.368
+ },
+ {
+ "epoch": 0.007100591715976331,
+ "grad_norm": 1.3350757360458374,
+ "learning_rate": 0.00017621621621621622,
+ "loss": 0.5705698132514954,
+ "num_input_tokens_seen": 28648,
+ "step": 30,
+ "train_runtime": 155.8918,
+ "train_tokens_per_second": 183.768
+ },
+ {
+ "epoch": 0.0073372781065088755,
+ "grad_norm": 0.68421471118927,
+ "learning_rate": 0.00017513513513513516,
+ "loss": 0.5660746097564697,
+ "num_input_tokens_seen": 29622,
+ "step": 31,
+ "train_runtime": 163.3209,
+ "train_tokens_per_second": 181.373
+ },
+ {
+ "epoch": 0.00757396449704142,
+ "grad_norm": 0.5762604475021362,
+ "learning_rate": 0.00017405405405405405,
+ "loss": 0.5483309030532837,
+ "num_input_tokens_seen": 30534,
+ "step": 32,
+ "train_runtime": 166.5185,
+ "train_tokens_per_second": 183.367
+ },
+ {
+ "epoch": 0.0078106508875739646,
+ "grad_norm": 0.5751784443855286,
+ "learning_rate": 0.000172972972972973,
+ "loss": 0.5207005739212036,
+ "num_input_tokens_seen": 31458,
+ "step": 33,
+ "train_runtime": 169.7055,
+ "train_tokens_per_second": 185.368
+ },
+ {
+ "epoch": 0.008047337278106508,
+ "grad_norm": 0.5752039551734924,
+ "learning_rate": 0.0001718918918918919,
+ "loss": 0.5534584522247314,
+ "num_input_tokens_seen": 32424,
+ "step": 34,
+ "train_runtime": 172.9184,
+ "train_tokens_per_second": 187.51
+ },
+ {
+ "epoch": 0.008284023668639054,
+ "grad_norm": 0.5918363332748413,
+ "learning_rate": 0.00017081081081081083,
+ "loss": 0.482683002948761,
+ "num_input_tokens_seen": 33398,
+ "step": 35,
+ "train_runtime": 176.2158,
+ "train_tokens_per_second": 189.529
+ },
+ {
+ "epoch": 0.008520710059171597,
+ "grad_norm": 0.6921846270561218,
+ "learning_rate": 0.00016972972972972974,
+ "loss": 0.46112507581710815,
+ "num_input_tokens_seen": 34324,
+ "step": 36,
+ "train_runtime": 179.3173,
+ "train_tokens_per_second": 191.415
+ },
+ {
+ "epoch": 0.008757396449704143,
+ "grad_norm": 0.6429721117019653,
+ "learning_rate": 0.00016864864864864866,
+ "loss": 0.45629197359085083,
+ "num_input_tokens_seen": 35270,
+ "step": 37,
+ "train_runtime": 182.3558,
+ "train_tokens_per_second": 193.413
+ },
+ {
+ "epoch": 0.008994082840236686,
+ "grad_norm": 0.5477020740509033,
+ "learning_rate": 0.00016756756756756757,
+ "loss": 0.4547877907752991,
+ "num_input_tokens_seen": 36206,
+ "step": 38,
+ "train_runtime": 185.4483,
+ "train_tokens_per_second": 195.235
+ },
+ {
+ "epoch": 0.009230769230769232,
+ "grad_norm": 0.6158634424209595,
+ "learning_rate": 0.00016648648648648652,
+ "loss": 0.367482990026474,
+ "num_input_tokens_seen": 37080,
+ "step": 39,
+ "train_runtime": 188.5114,
+ "train_tokens_per_second": 196.699
+ },
+ {
+ "epoch": 0.009467455621301775,
+ "grad_norm": 0.5405091643333435,
+ "learning_rate": 0.0001654054054054054,
+ "loss": 0.42049410939216614,
+ "num_input_tokens_seen": 38030,
+ "step": 40,
+ "train_runtime": 191.5986,
+ "train_tokens_per_second": 198.488
+ },
+ {
+ "epoch": 0.009704142011834319,
+ "grad_norm": 0.6389183402061462,
+ "learning_rate": 0.00016432432432432435,
+ "loss": 0.3935014307498932,
+ "num_input_tokens_seen": 38942,
+ "step": 41,
+ "train_runtime": 194.8426,
+ "train_tokens_per_second": 199.864
+ },
+ {
+ "epoch": 0.009940828402366864,
+ "grad_norm": 0.6742956042289734,
+ "learning_rate": 0.00016324324324324324,
+ "loss": 0.4373692274093628,
+ "num_input_tokens_seen": 39902,
+ "step": 42,
+ "train_runtime": 197.8938,
+ "train_tokens_per_second": 201.633
+ },
+ {
+ "epoch": 0.010177514792899408,
+ "grad_norm": 0.6083794236183167,
+ "learning_rate": 0.00016216216216216218,
+ "loss": 0.3776765465736389,
+ "num_input_tokens_seen": 40858,
+ "step": 43,
+ "train_runtime": 200.9708,
+ "train_tokens_per_second": 203.303
+ },
+ {
+ "epoch": 0.010414201183431953,
+ "grad_norm": 0.5469992160797119,
+ "learning_rate": 0.0001610810810810811,
+ "loss": 0.31133151054382324,
+ "num_input_tokens_seen": 41760,
+ "step": 44,
+ "train_runtime": 204.1304,
+ "train_tokens_per_second": 204.575
+ },
+ {
+ "epoch": 0.010650887573964497,
+ "grad_norm": 0.6452498435974121,
+ "learning_rate": 0.00016,
+ "loss": 0.42721521854400635,
+ "num_input_tokens_seen": 42740,
+ "step": 45,
+ "train_runtime": 207.1845,
+ "train_tokens_per_second": 206.29
+ },
+ {
+ "epoch": 0.01088757396449704,
+ "grad_norm": 0.5988945960998535,
+ "learning_rate": 0.00015891891891891893,
+ "loss": 0.41375258564949036,
+ "num_input_tokens_seen": 43750,
+ "step": 46,
+ "train_runtime": 210.2912,
+ "train_tokens_per_second": 208.045
+ },
+ {
+ "epoch": 0.011124260355029586,
+ "grad_norm": 0.6452038288116455,
+ "learning_rate": 0.00015783783783783785,
+ "loss": 0.3485276699066162,
+ "num_input_tokens_seen": 44690,
+ "step": 47,
+ "train_runtime": 213.3684,
+ "train_tokens_per_second": 209.45
+ },
+ {
+ "epoch": 0.01136094674556213,
+ "grad_norm": 0.7041454911231995,
+ "learning_rate": 0.00015675675675675676,
+ "loss": 0.3296637535095215,
+ "num_input_tokens_seen": 45666,
+ "step": 48,
+ "train_runtime": 216.4459,
+ "train_tokens_per_second": 210.981
+ },
+ {
+ "epoch": 0.011597633136094675,
+ "grad_norm": 0.6726324558258057,
+ "learning_rate": 0.0001556756756756757,
+ "loss": 0.3179457187652588,
+ "num_input_tokens_seen": 46560,
+ "step": 49,
+ "train_runtime": 219.4697,
+ "train_tokens_per_second": 212.148
+ },
+ {
+ "epoch": 0.011834319526627219,
+ "grad_norm": 0.7513703107833862,
+ "learning_rate": 0.0001545945945945946,
+ "loss": 0.3722391426563263,
+ "num_input_tokens_seen": 47528,
+ "step": 50,
+ "train_runtime": 222.596,
+ "train_tokens_per_second": 213.517
+ },
+ {
+ "epoch": 0.012071005917159764,
+ "grad_norm": 0.6454160213470459,
+ "learning_rate": 0.00015351351351351354,
+ "loss": 0.283376544713974,
+ "num_input_tokens_seen": 48468,
+ "step": 51,
+ "train_runtime": 225.7262,
+ "train_tokens_per_second": 214.72
+ },
+ {
+ "epoch": 0.012307692307692308,
+ "grad_norm": 0.9116244316101074,
+ "learning_rate": 0.00015243243243243243,
+ "loss": 0.4286752939224243,
+ "num_input_tokens_seen": 49444,
+ "step": 52,
+ "train_runtime": 228.803,
+ "train_tokens_per_second": 216.099
+ },
+ {
+ "epoch": 0.012544378698224851,
+ "grad_norm": 0.6157099604606628,
+ "learning_rate": 0.00015135135135135137,
+ "loss": 0.27355721592903137,
+ "num_input_tokens_seen": 50424,
+ "step": 53,
+ "train_runtime": 231.8413,
+ "train_tokens_per_second": 217.494
+ },
+ {
+ "epoch": 0.012781065088757397,
+ "grad_norm": 0.6335808038711548,
+ "learning_rate": 0.00015027027027027028,
+ "loss": 0.37197017669677734,
+ "num_input_tokens_seen": 51328,
+ "step": 54,
+ "train_runtime": 234.9265,
+ "train_tokens_per_second": 218.485
+ },
+ {
+ "epoch": 0.01301775147928994,
+ "grad_norm": 0.6838424205780029,
+ "learning_rate": 0.0001491891891891892,
+ "loss": 0.3504602909088135,
+ "num_input_tokens_seen": 52274,
+ "step": 55,
+ "train_runtime": 238.0114,
+ "train_tokens_per_second": 219.628
+ },
+ {
+ "epoch": 0.013254437869822486,
+ "grad_norm": 0.6566418409347534,
+ "learning_rate": 0.00014810810810810812,
+ "loss": 0.3189927935600281,
+ "num_input_tokens_seen": 53214,
+ "step": 56,
+ "train_runtime": 241.0892,
+ "train_tokens_per_second": 220.723
+ },
+ {
+ "epoch": 0.01349112426035503,
+ "grad_norm": 0.6112971901893616,
+ "learning_rate": 0.00014702702702702703,
+ "loss": 0.3080369830131531,
+ "num_input_tokens_seen": 54162,
+ "step": 57,
+ "train_runtime": 244.2461,
+ "train_tokens_per_second": 221.752
+ },
+ {
+ "epoch": 0.013727810650887575,
+ "grad_norm": 0.6469757556915283,
+ "learning_rate": 0.00014594594594594595,
+ "loss": 0.24279898405075073,
+ "num_input_tokens_seen": 55158,
+ "step": 58,
+ "train_runtime": 247.4064,
+ "train_tokens_per_second": 222.945
+ },
+ {
+ "epoch": 0.013964497041420118,
+ "grad_norm": 0.7099384665489197,
+ "learning_rate": 0.0001448648648648649,
+ "loss": 0.31188705563545227,
+ "num_input_tokens_seen": 56096,
+ "step": 59,
+ "train_runtime": 250.4708,
+ "train_tokens_per_second": 223.962
+ },
+ {
+ "epoch": 0.014201183431952662,
+ "grad_norm": 0.5927040576934814,
+ "learning_rate": 0.00014378378378378378,
+ "loss": 0.25696688890457153,
+ "num_input_tokens_seen": 57012,
+ "step": 60,
+ "train_runtime": 253.7131,
+ "train_tokens_per_second": 224.711
+ },
+ {
+ "epoch": 0.014437869822485207,
+ "grad_norm": 0.5483338236808777,
+ "learning_rate": 0.00014270270270270272,
+ "loss": 0.2725352644920349,
+ "num_input_tokens_seen": 58036,
+ "step": 61,
+ "train_runtime": 260.2844,
+ "train_tokens_per_second": 222.971
+ },
+ {
+ "epoch": 0.014674556213017751,
+ "grad_norm": 0.6806847453117371,
+ "learning_rate": 0.00014162162162162161,
+ "loss": 0.31959983706474304,
+ "num_input_tokens_seen": 59026,
+ "step": 62,
+ "train_runtime": 263.3802,
+ "train_tokens_per_second": 224.109
+ },
+ {
+ "epoch": 0.014911242603550296,
+ "grad_norm": 0.6916660070419312,
+ "learning_rate": 0.00014054054054054056,
+ "loss": 0.312305748462677,
+ "num_input_tokens_seen": 59982,
+ "step": 63,
+ "train_runtime": 266.3794,
+ "train_tokens_per_second": 225.175
+ },
+ {
+ "epoch": 0.01514792899408284,
+ "grad_norm": 0.5516266226768494,
+ "learning_rate": 0.00013945945945945947,
+ "loss": 0.2653953433036804,
+ "num_input_tokens_seen": 60926,
+ "step": 64,
+ "train_runtime": 269.3893,
+ "train_tokens_per_second": 226.163
+ },
+ {
+ "epoch": 0.015384615384615385,
+ "grad_norm": 0.5579270720481873,
+ "learning_rate": 0.0001383783783783784,
+ "loss": 0.20462818443775177,
+ "num_input_tokens_seen": 61844,
+ "step": 65,
+ "train_runtime": 272.3982,
+ "train_tokens_per_second": 227.035
+ },
+ {
+ "epoch": 0.015621301775147929,
+ "grad_norm": 0.7645660638809204,
+ "learning_rate": 0.0001372972972972973,
+ "loss": 0.2650893032550812,
+ "num_input_tokens_seen": 62766,
+ "step": 66,
+ "train_runtime": 275.404,
+ "train_tokens_per_second": 227.905
+ },
+ {
+ "epoch": 0.015857988165680473,
+ "grad_norm": 0.5974738001823425,
+ "learning_rate": 0.00013621621621621622,
+ "loss": 0.21247778832912445,
+ "num_input_tokens_seen": 63690,
+ "step": 67,
+ "train_runtime": 278.4021,
+ "train_tokens_per_second": 228.77
+ },
+ {
+ "epoch": 0.016094674556213016,
+ "grad_norm": 0.5831193327903748,
+ "learning_rate": 0.00013513513513513514,
+ "loss": 0.2024172842502594,
+ "num_input_tokens_seen": 64496,
+ "step": 68,
+ "train_runtime": 281.4313,
+ "train_tokens_per_second": 229.171
+ },
+ {
+ "epoch": 0.016331360946745564,
+ "grad_norm": 0.8678795695304871,
+ "learning_rate": 0.00013405405405405408,
+ "loss": 0.31490638852119446,
+ "num_input_tokens_seen": 65460,
+ "step": 69,
+ "train_runtime": 284.6137,
+ "train_tokens_per_second": 229.996
+ },
+ {
+ "epoch": 0.016568047337278107,
+ "grad_norm": 0.7611344456672668,
+ "learning_rate": 0.00013297297297297297,
+ "loss": 0.24661925435066223,
+ "num_input_tokens_seen": 66374,
+ "step": 70,
+ "train_runtime": 287.7742,
+ "train_tokens_per_second": 230.646
+ },
+ {
+ "epoch": 0.01680473372781065,
+ "grad_norm": 0.6685308814048767,
+ "learning_rate": 0.0001318918918918919,
+ "loss": 0.1843535602092743,
+ "num_input_tokens_seen": 67112,
+ "step": 71,
+ "train_runtime": 290.9321,
+ "train_tokens_per_second": 230.679
+ },
+ {
+ "epoch": 0.017041420118343194,
+ "grad_norm": 0.6150883436203003,
+ "learning_rate": 0.0001308108108108108,
+ "loss": 0.23372295498847961,
+ "num_input_tokens_seen": 68154,
+ "step": 72,
+ "train_runtime": 294.0809,
+ "train_tokens_per_second": 231.753
+ },
+ {
+ "epoch": 0.017278106508875738,
+ "grad_norm": 0.6938812732696533,
+ "learning_rate": 0.00012972972972972974,
+ "loss": 0.22619186341762543,
+ "num_input_tokens_seen": 69102,
+ "step": 73,
+ "train_runtime": 297.1843,
+ "train_tokens_per_second": 232.522
+ },
+ {
+ "epoch": 0.017514792899408285,
+ "grad_norm": 0.617787778377533,
+ "learning_rate": 0.00012864864864864866,
+ "loss": 0.2535328269004822,
+ "num_input_tokens_seen": 70010,
+ "step": 74,
+ "train_runtime": 300.2851,
+ "train_tokens_per_second": 233.145
+ },
+ {
+ "epoch": 0.01775147928994083,
+ "grad_norm": 0.6884900331497192,
+ "learning_rate": 0.00012756756756756758,
+ "loss": 0.21632739901542664,
+ "num_input_tokens_seen": 70940,
+ "step": 75,
+ "train_runtime": 303.3533,
+ "train_tokens_per_second": 233.853
+ },
+ {
+ "epoch": 0.017988165680473372,
+ "grad_norm": 0.5782189965248108,
+ "learning_rate": 0.0001264864864864865,
+ "loss": 0.20167675614356995,
+ "num_input_tokens_seen": 71862,
+ "step": 76,
+ "train_runtime": 306.4503,
+ "train_tokens_per_second": 234.498
+ },
+ {
+ "epoch": 0.018224852071005916,
+ "grad_norm": 0.6541483402252197,
+ "learning_rate": 0.0001254054054054054,
+ "loss": 0.23231221735477448,
+ "num_input_tokens_seen": 72812,
+ "step": 77,
+ "train_runtime": 309.5743,
+ "train_tokens_per_second": 235.2
+ },
+ {
+ "epoch": 0.018461538461538463,
+ "grad_norm": 0.7524886727333069,
+ "learning_rate": 0.00012432432432432433,
+ "loss": 0.27632662653923035,
+ "num_input_tokens_seen": 73774,
+ "step": 78,
+ "train_runtime": 312.6963,
+ "train_tokens_per_second": 235.929
+ },
+ {
+ "epoch": 0.018698224852071007,
+ "grad_norm": 0.716282308101654,
+ "learning_rate": 0.00012324324324324327,
+ "loss": 0.2734939754009247,
+ "num_input_tokens_seen": 74716,
+ "step": 79,
+ "train_runtime": 315.7612,
+ "train_tokens_per_second": 236.622
+ },
+ {
+ "epoch": 0.01893491124260355,
+ "grad_norm": 0.7010631561279297,
+ "learning_rate": 0.00012216216216216216,
+ "loss": 0.2774362564086914,
+ "num_input_tokens_seen": 75692,
+ "step": 80,
+ "train_runtime": 318.7983,
+ "train_tokens_per_second": 237.429
+ },
+ {
+ "epoch": 0.019171597633136094,
+ "grad_norm": 0.6561240553855896,
+ "learning_rate": 0.00012108108108108109,
+ "loss": 0.21640686690807343,
+ "num_input_tokens_seen": 76558,
+ "step": 81,
+ "train_runtime": 321.889,
+ "train_tokens_per_second": 237.84
+ },
+ {
+ "epoch": 0.019408284023668638,
+ "grad_norm": 0.5552831888198853,
+ "learning_rate": 0.00012,
+ "loss": 0.21232956647872925,
+ "num_input_tokens_seen": 77408,
+ "step": 82,
+ "train_runtime": 325.0469,
+ "train_tokens_per_second": 238.144
+ },
+ {
+ "epoch": 0.019644970414201185,
+ "grad_norm": 0.5905120372772217,
+ "learning_rate": 0.00011891891891891893,
+ "loss": 0.22443550825119019,
+ "num_input_tokens_seen": 78278,
+ "step": 83,
+ "train_runtime": 328.099,
+ "train_tokens_per_second": 238.58
+ },
+ {
+ "epoch": 0.01988165680473373,
+ "grad_norm": 0.6606366634368896,
+ "learning_rate": 0.00011783783783783784,
+ "loss": 0.2400747388601303,
+ "num_input_tokens_seen": 79262,
+ "step": 84,
+ "train_runtime": 331.1361,
+ "train_tokens_per_second": 239.364
+ },
+ {
+ "epoch": 0.020118343195266272,
+ "grad_norm": 0.6508055329322815,
+ "learning_rate": 0.00011675675675675676,
+ "loss": 0.19072499871253967,
+ "num_input_tokens_seen": 80144,
+ "step": 85,
+ "train_runtime": 334.2158,
+ "train_tokens_per_second": 239.797
+ },
+ {
+ "epoch": 0.020355029585798816,
+ "grad_norm": 0.606562614440918,
+ "learning_rate": 0.00011567567567567568,
+ "loss": 0.20951862633228302,
+ "num_input_tokens_seen": 81108,
+ "step": 86,
+ "train_runtime": 337.3105,
+ "train_tokens_per_second": 240.455
+ },
+ {
+ "epoch": 0.02059171597633136,
+ "grad_norm": 0.6757521629333496,
+ "learning_rate": 0.00011459459459459461,
+ "loss": 0.17614495754241943,
+ "num_input_tokens_seen": 82032,
+ "step": 87,
+ "train_runtime": 340.3438,
+ "train_tokens_per_second": 241.027
+ },
+ {
+ "epoch": 0.020828402366863907,
+ "grad_norm": 0.671481728553772,
+ "learning_rate": 0.00011351351351351351,
+ "loss": 0.1901710331439972,
+ "num_input_tokens_seen": 82954,
+ "step": 88,
+ "train_runtime": 343.4059,
+ "train_tokens_per_second": 241.563
+ },
+ {
+ "epoch": 0.02106508875739645,
+ "grad_norm": 0.5919027924537659,
+ "learning_rate": 0.00011243243243243244,
+ "loss": 0.1563737988471985,
+ "num_input_tokens_seen": 83880,
+ "step": 89,
+ "train_runtime": 346.5125,
+ "train_tokens_per_second": 242.069
+ },
+ {
+ "epoch": 0.021301775147928994,
+ "grad_norm": 0.7108309268951416,
+ "learning_rate": 0.00011135135135135135,
+ "loss": 0.17256894707679749,
+ "num_input_tokens_seen": 84852,
+ "step": 90,
+ "train_runtime": 349.6508,
+ "train_tokens_per_second": 242.676
+ },
+ {
+ "epoch": 0.021538461538461538,
+ "grad_norm": 0.8608420491218567,
+ "learning_rate": 0.00011027027027027029,
+ "loss": 0.1845530867576599,
+ "num_input_tokens_seen": 85820,
+ "step": 91,
+ "train_runtime": 356.267,
+ "train_tokens_per_second": 240.887
+ },
+ {
+ "epoch": 0.02177514792899408,
+ "grad_norm": 0.9677010774612427,
+ "learning_rate": 0.00010918918918918919,
+ "loss": 0.24358105659484863,
+ "num_input_tokens_seen": 86770,
+ "step": 92,
+ "train_runtime": 359.3722,
+ "train_tokens_per_second": 241.449
+ },
+ {
+ "epoch": 0.02201183431952663,
+ "grad_norm": 0.6302966475486755,
+ "learning_rate": 0.00010810810810810812,
+ "loss": 0.18155133724212646,
+ "num_input_tokens_seen": 87736,
+ "step": 93,
+ "train_runtime": 362.5874,
+ "train_tokens_per_second": 241.972
+ },
+ {
+ "epoch": 0.022248520710059172,
+ "grad_norm": 0.7171812653541565,
+ "learning_rate": 0.00010702702702702702,
+ "loss": 0.15349116921424866,
+ "num_input_tokens_seen": 88626,
+ "step": 94,
+ "train_runtime": 365.7055,
+ "train_tokens_per_second": 242.343
+ },
+ {
+ "epoch": 0.022485207100591716,
+ "grad_norm": 0.7326887845993042,
+ "learning_rate": 0.00010594594594594595,
+ "loss": 0.18243758380413055,
+ "num_input_tokens_seen": 89548,
+ "step": 95,
+ "train_runtime": 368.799,
+ "train_tokens_per_second": 242.81
+ },
+ {
+ "epoch": 0.02272189349112426,
+ "grad_norm": 0.6686019897460938,
+ "learning_rate": 0.00010486486486486487,
+ "loss": 0.15622465312480927,
+ "num_input_tokens_seen": 90504,
+ "step": 96,
+ "train_runtime": 371.8404,
+ "train_tokens_per_second": 243.395
+ },
+ {
+ "epoch": 0.022958579881656806,
+ "grad_norm": 0.6306503415107727,
+ "learning_rate": 0.0001037837837837838,
+ "loss": 0.1272117793560028,
+ "num_input_tokens_seen": 91454,
+ "step": 97,
+ "train_runtime": 375.0199,
+ "train_tokens_per_second": 243.864
+ },
+ {
+ "epoch": 0.02319526627218935,
+ "grad_norm": 0.6068493127822876,
+ "learning_rate": 0.0001027027027027027,
+ "loss": 0.1577155441045761,
+ "num_input_tokens_seen": 92466,
+ "step": 98,
+ "train_runtime": 378.0978,
+ "train_tokens_per_second": 244.556
+ },
+ {
+ "epoch": 0.023431952662721894,
+ "grad_norm": 0.7263635993003845,
+ "learning_rate": 0.00010162162162162163,
+ "loss": 0.1845340132713318,
+ "num_input_tokens_seen": 93398,
+ "step": 99,
+ "train_runtime": 381.1358,
+ "train_tokens_per_second": 245.052
+ },
+ {
+ "epoch": 0.023668639053254437,
+ "grad_norm": 0.6466794013977051,
+ "learning_rate": 0.00010054054054054053,
+ "loss": 0.14443959295749664,
+ "num_input_tokens_seen": 94230,
+ "step": 100,
+ "train_runtime": 384.2906,
+ "train_tokens_per_second": 245.205
+ },
+ {
+ "epoch": 0.02390532544378698,
+ "grad_norm": 0.6924043297767639,
+ "learning_rate": 9.945945945945948e-05,
+ "loss": 0.18473166227340698,
+ "num_input_tokens_seen": 95132,
+ "step": 101,
+ "train_runtime": 387.5005,
+ "train_tokens_per_second": 245.502
+ },
+ {
+ "epoch": 0.024142011834319528,
+ "grad_norm": 0.732337236404419,
+ "learning_rate": 9.837837837837839e-05,
+ "loss": 0.16353662312030792,
+ "num_input_tokens_seen": 96080,
+ "step": 102,
+ "train_runtime": 390.584,
+ "train_tokens_per_second": 245.991
+ },
+ {
+ "epoch": 0.02437869822485207,
+ "grad_norm": 0.5956932902336121,
+ "learning_rate": 9.729729729729731e-05,
+ "loss": 0.16359847784042358,
+ "num_input_tokens_seen": 97048,
+ "step": 103,
+ "train_runtime": 393.6474,
+ "train_tokens_per_second": 246.535
+ },
+ {
+ "epoch": 0.024615384615384615,
+ "grad_norm": 0.6137388348579407,
+ "learning_rate": 9.621621621621622e-05,
+ "loss": 0.15779417753219604,
+ "num_input_tokens_seen": 97986,
+ "step": 104,
+ "train_runtime": 396.7042,
+ "train_tokens_per_second": 247.0
+ },
+ {
+ "epoch": 0.02485207100591716,
+ "grad_norm": 0.7936599254608154,
+ "learning_rate": 9.513513513513514e-05,
+ "loss": 0.21357379853725433,
+ "num_input_tokens_seen": 98894,
+ "step": 105,
+ "train_runtime": 399.8228,
+ "train_tokens_per_second": 247.345
+ },
+ {
+ "epoch": 0.025088757396449703,
+ "grad_norm": 0.6574507355690002,
+ "learning_rate": 9.405405405405407e-05,
+ "loss": 0.1685573309659958,
+ "num_input_tokens_seen": 99856,
+ "step": 106,
+ "train_runtime": 402.8867,
+ "train_tokens_per_second": 247.851
+ },
+ {
+ "epoch": 0.02532544378698225,
+ "grad_norm": 0.5891634821891785,
+ "learning_rate": 9.297297297297299e-05,
+ "loss": 0.12082140147686005,
+ "num_input_tokens_seen": 100810,
+ "step": 107,
+ "train_runtime": 405.9752,
+ "train_tokens_per_second": 248.316
+ },
+ {
+ "epoch": 0.025562130177514793,
+ "grad_norm": 0.6624342799186707,
+ "learning_rate": 9.18918918918919e-05,
+ "loss": 0.1473795771598816,
+ "num_input_tokens_seen": 101744,
+ "step": 108,
+ "train_runtime": 409.0084,
+ "train_tokens_per_second": 248.758
+ },
+ {
+ "epoch": 0.025798816568047337,
+ "grad_norm": 0.5968036651611328,
+ "learning_rate": 9.081081081081082e-05,
+ "loss": 0.12923751771450043,
+ "num_input_tokens_seen": 102702,
+ "step": 109,
+ "train_runtime": 412.199,
+ "train_tokens_per_second": 249.156
+ },
+ {
+ "epoch": 0.02603550295857988,
+ "grad_norm": 0.5954580307006836,
+ "learning_rate": 8.972972972972973e-05,
+ "loss": 0.1161399856209755,
+ "num_input_tokens_seen": 103672,
+ "step": 110,
+ "train_runtime": 415.3736,
+ "train_tokens_per_second": 249.587
+ },
+ {
+ "epoch": 0.026272189349112424,
+ "grad_norm": 0.6139379739761353,
+ "learning_rate": 8.864864864864866e-05,
+ "loss": 0.12352899461984634,
+ "num_input_tokens_seen": 104598,
+ "step": 111,
+ "train_runtime": 418.3271,
+ "train_tokens_per_second": 250.039
+ },
+ {
+ "epoch": 0.02650887573964497,
+ "grad_norm": 0.7933095097541809,
+ "learning_rate": 8.756756756756758e-05,
+ "loss": 0.1709994524717331,
+ "num_input_tokens_seen": 105580,
+ "step": 112,
+ "train_runtime": 421.3565,
+ "train_tokens_per_second": 250.572
+ },
+ {
+ "epoch": 0.026745562130177515,
+ "grad_norm": 0.7641370892524719,
+ "learning_rate": 8.64864864864865e-05,
+ "loss": 0.13086311519145966,
+ "num_input_tokens_seen": 106444,
+ "step": 113,
+ "train_runtime": 424.4668,
+ "train_tokens_per_second": 250.771
+ },
+ {
+ "epoch": 0.02698224852071006,
+ "grad_norm": 0.7611938118934631,
+ "learning_rate": 8.540540540540541e-05,
+ "loss": 0.16537640988826752,
+ "num_input_tokens_seen": 107454,
+ "step": 114,
+ "train_runtime": 427.5,
+ "train_tokens_per_second": 251.354
+ },
+ {
+ "epoch": 0.027218934911242602,
+ "grad_norm": 0.7495304346084595,
+ "learning_rate": 8.432432432432433e-05,
+ "loss": 0.14228317141532898,
+ "num_input_tokens_seen": 108426,
+ "step": 115,
+ "train_runtime": 430.5035,
+ "train_tokens_per_second": 251.859
+ },
+ {
+ "epoch": 0.02745562130177515,
+ "grad_norm": 1.2078282833099365,
+ "learning_rate": 8.324324324324326e-05,
+ "loss": 0.10530930012464523,
+ "num_input_tokens_seen": 109446,
+ "step": 116,
+ "train_runtime": 433.4964,
+ "train_tokens_per_second": 252.473
+ },
+ {
+ "epoch": 0.027692307692307693,
+ "grad_norm": 0.6765572428703308,
+ "learning_rate": 8.216216216216217e-05,
+ "loss": 0.10717278718948364,
+ "num_input_tokens_seen": 110374,
+ "step": 117,
+ "train_runtime": 436.5901,
+ "train_tokens_per_second": 252.809
+ },
+ {
+ "epoch": 0.027928994082840237,
+ "grad_norm": 0.8501551151275635,
+ "learning_rate": 8.108108108108109e-05,
+ "loss": 0.17784273624420166,
+ "num_input_tokens_seen": 111340,
+ "step": 118,
+ "train_runtime": 439.6526,
+ "train_tokens_per_second": 253.245
+ },
+ {
+ "epoch": 0.02816568047337278,
+ "grad_norm": 0.7583926320075989,
+ "learning_rate": 8e-05,
+ "loss": 0.16588062047958374,
+ "num_input_tokens_seen": 112288,
+ "step": 119,
+ "train_runtime": 442.8046,
+ "train_tokens_per_second": 253.584
+ },
+ {
+ "epoch": 0.028402366863905324,
+ "grad_norm": 0.6620173454284668,
+ "learning_rate": 7.891891891891892e-05,
+ "loss": 0.11676295101642609,
+ "num_input_tokens_seen": 113206,
+ "step": 120,
+ "train_runtime": 445.9682,
+ "train_tokens_per_second": 253.843
+ },
+ {
+ "epoch": 0.02863905325443787,
+ "grad_norm": 0.6844660639762878,
+ "learning_rate": 7.783783783783785e-05,
+ "loss": 0.09329869598150253,
+ "num_input_tokens_seen": 114134,
+ "step": 121,
+ "train_runtime": 452.5692,
+ "train_tokens_per_second": 252.191
+ },
+ {
+ "epoch": 0.028875739644970415,
+ "grad_norm": 0.7968376874923706,
+ "learning_rate": 7.675675675675677e-05,
+ "loss": 0.13320565223693848,
+ "num_input_tokens_seen": 115088,
+ "step": 122,
+ "train_runtime": 455.6245,
+ "train_tokens_per_second": 252.594
+ },
+ {
+ "epoch": 0.02911242603550296,
+ "grad_norm": 0.8079301118850708,
+ "learning_rate": 7.567567567567568e-05,
+ "loss": 0.1387016326189041,
+ "num_input_tokens_seen": 116006,
+ "step": 123,
+ "train_runtime": 458.7601,
+ "train_tokens_per_second": 252.869
+ },
+ {
+ "epoch": 0.029349112426035502,
+ "grad_norm": 1.1301748752593994,
+ "learning_rate": 7.45945945945946e-05,
+ "loss": 0.197790265083313,
+ "num_input_tokens_seen": 116958,
+ "step": 124,
+ "train_runtime": 461.861,
+ "train_tokens_per_second": 253.232
+ },
+ {
+ "epoch": 0.029585798816568046,
+ "grad_norm": 0.7272042632102966,
+ "learning_rate": 7.351351351351352e-05,
+ "loss": 0.12121891975402832,
+ "num_input_tokens_seen": 117886,
+ "step": 125,
+ "train_runtime": 465.0328,
+ "train_tokens_per_second": 253.5
+ },
+ {
+ "epoch": 0.029822485207100593,
+ "grad_norm": 0.6021063923835754,
+ "learning_rate": 7.243243243243245e-05,
+ "loss": 0.08234579861164093,
+ "num_input_tokens_seen": 118806,
+ "step": 126,
+ "train_runtime": 468.1518,
+ "train_tokens_per_second": 253.777
+ },
+ {
+ "epoch": 0.030059171597633137,
+ "grad_norm": 0.5659137964248657,
+ "learning_rate": 7.135135135135136e-05,
+ "loss": 0.07469362020492554,
+ "num_input_tokens_seen": 119716,
+ "step": 127,
+ "train_runtime": 471.2559,
+ "train_tokens_per_second": 254.036
+ },
+ {
+ "epoch": 0.03029585798816568,
+ "grad_norm": 0.7604466080665588,
+ "learning_rate": 7.027027027027028e-05,
+ "loss": 0.15167035162448883,
+ "num_input_tokens_seen": 120710,
+ "step": 128,
+ "train_runtime": 474.3536,
+ "train_tokens_per_second": 254.473
+ },
+ {
+ "epoch": 0.030532544378698224,
+ "grad_norm": 0.898281455039978,
+ "learning_rate": 6.91891891891892e-05,
+ "loss": 0.18358615040779114,
+ "num_input_tokens_seen": 121668,
+ "step": 129,
+ "train_runtime": 477.4676,
+ "train_tokens_per_second": 254.819
+ },
+ {
+ "epoch": 0.03076923076923077,
+ "grad_norm": 1.046539545059204,
+ "learning_rate": 6.810810810810811e-05,
+ "loss": 0.14173445105552673,
+ "num_input_tokens_seen": 122654,
+ "step": 130,
+ "train_runtime": 480.6242,
+ "train_tokens_per_second": 255.197
+ },
+ {
+ "epoch": 0.031005917159763315,
+ "grad_norm": 0.689860463142395,
+ "learning_rate": 6.702702702702704e-05,
+ "loss": 0.1232977956533432,
+ "num_input_tokens_seen": 123610,
+ "step": 131,
+ "train_runtime": 483.7904,
+ "train_tokens_per_second": 255.503
+ },
+ {
+ "epoch": 0.031242603550295858,
+ "grad_norm": 0.6417210698127747,
+ "learning_rate": 6.594594594594596e-05,
+ "loss": 0.08111131936311722,
+ "num_input_tokens_seen": 124526,
+ "step": 132,
+ "train_runtime": 486.9241,
+ "train_tokens_per_second": 255.74
+ },
+ {
+ "epoch": 0.0314792899408284,
+ "grad_norm": 0.8019670844078064,
+ "learning_rate": 6.486486486486487e-05,
+ "loss": 0.13235916197299957,
+ "num_input_tokens_seen": 125438,
+ "step": 133,
+ "train_runtime": 490.0416,
+ "train_tokens_per_second": 255.974
+ },
+ {
+ "epoch": 0.031715976331360946,
+ "grad_norm": 0.8618372082710266,
+ "learning_rate": 6.378378378378379e-05,
+ "loss": 0.11209192872047424,
+ "num_input_tokens_seen": 126378,
+ "step": 134,
+ "train_runtime": 493.0816,
+ "train_tokens_per_second": 256.302
+ },
+ {
+ "epoch": 0.03195266272189349,
+ "grad_norm": 0.6566123962402344,
+ "learning_rate": 6.27027027027027e-05,
+ "loss": 0.08107412606477737,
+ "num_input_tokens_seen": 127280,
+ "step": 135,
+ "train_runtime": 496.1373,
+ "train_tokens_per_second": 256.542
+ },
+ {
+ "epoch": 0.03218934911242603,
+ "grad_norm": 0.671886682510376,
+ "learning_rate": 6.162162162162163e-05,
+ "loss": 0.10395485162734985,
+ "num_input_tokens_seen": 128246,
+ "step": 136,
+ "train_runtime": 499.2198,
+ "train_tokens_per_second": 256.893
+ },
+ {
+ "epoch": 0.032426035502958576,
+ "grad_norm": 0.5853029489517212,
+ "learning_rate": 6.0540540540540543e-05,
+ "loss": 0.06830425560474396,
+ "num_input_tokens_seen": 129142,
+ "step": 137,
+ "train_runtime": 502.2719,
+ "train_tokens_per_second": 257.116
+ },
+ {
+ "epoch": 0.03266272189349113,
+ "grad_norm": 0.9871621131896973,
+ "learning_rate": 5.9459459459459466e-05,
+ "loss": 0.14138084650039673,
+ "num_input_tokens_seen": 130060,
+ "step": 138,
+ "train_runtime": 505.3687,
+ "train_tokens_per_second": 257.357
+ },
+ {
+ "epoch": 0.03289940828402367,
+ "grad_norm": 0.6716732978820801,
+ "learning_rate": 5.837837837837838e-05,
+ "loss": 0.07525799423456192,
+ "num_input_tokens_seen": 130982,
+ "step": 139,
+ "train_runtime": 508.4248,
+ "train_tokens_per_second": 257.623
+ },
+ {
+ "epoch": 0.033136094674556214,
+ "grad_norm": 0.7177789807319641,
+ "learning_rate": 5.7297297297297305e-05,
+ "loss": 0.09517940878868103,
+ "num_input_tokens_seen": 131900,
+ "step": 140,
+ "train_runtime": 511.57,
+ "train_tokens_per_second": 257.834
+ },
+ {
+ "epoch": 0.03337278106508876,
+ "grad_norm": 0.703551709651947,
+ "learning_rate": 5.621621621621622e-05,
+ "loss": 0.09361469000577927,
+ "num_input_tokens_seen": 132890,
+ "step": 141,
+ "train_runtime": 514.7186,
+ "train_tokens_per_second": 258.18
+ },
+ {
+ "epoch": 0.0336094674556213,
+ "grad_norm": 0.7156357765197754,
+ "learning_rate": 5.5135135135135144e-05,
+ "loss": 0.08832181245088577,
+ "num_input_tokens_seen": 133810,
+ "step": 142,
+ "train_runtime": 517.7644,
+ "train_tokens_per_second": 258.438
+ },
+ {
+ "epoch": 0.033846153846153845,
+ "grad_norm": 0.7757886648178101,
+ "learning_rate": 5.405405405405406e-05,
+ "loss": 0.12478049844503403,
+ "num_input_tokens_seen": 134758,
+ "step": 143,
+ "train_runtime": 520.8215,
+ "train_tokens_per_second": 258.741
+ },
+ {
+ "epoch": 0.03408284023668639,
+ "grad_norm": 0.6047410368919373,
+ "learning_rate": 5.2972972972972976e-05,
+ "loss": 0.07645335048437119,
+ "num_input_tokens_seen": 135724,
+ "step": 144,
+ "train_runtime": 523.9787,
+ "train_tokens_per_second": 259.026
+ },
+ {
+ "epoch": 0.03431952662721893,
+ "grad_norm": 0.7750188708305359,
+ "learning_rate": 5.18918918918919e-05,
+ "loss": 0.10904290527105331,
+ "num_input_tokens_seen": 136662,
+ "step": 145,
+ "train_runtime": 527.0633,
+ "train_tokens_per_second": 259.29
+ },
+ {
+ "epoch": 0.034556213017751476,
+ "grad_norm": 0.6400149464607239,
+ "learning_rate": 5.0810810810810815e-05,
+ "loss": 0.07525563985109329,
+ "num_input_tokens_seen": 137628,
+ "step": 146,
+ "train_runtime": 530.1299,
+ "train_tokens_per_second": 259.612
+ },
+ {
+ "epoch": 0.03479289940828403,
+ "grad_norm": 0.6681646108627319,
+ "learning_rate": 4.972972972972974e-05,
+ "loss": 0.08536849915981293,
+ "num_input_tokens_seen": 138610,
+ "step": 147,
+ "train_runtime": 533.159,
+ "train_tokens_per_second": 259.979
+ },
+ {
+ "epoch": 0.03502958579881657,
+ "grad_norm": 0.6158885359764099,
+ "learning_rate": 4.8648648648648654e-05,
+ "loss": 0.0753958523273468,
+ "num_input_tokens_seen": 139486,
+ "step": 148,
+ "train_runtime": 536.2632,
+ "train_tokens_per_second": 260.107
+ },
+ {
+ "epoch": 0.035266272189349114,
+ "grad_norm": 0.665565550327301,
+ "learning_rate": 4.756756756756757e-05,
+ "loss": 0.08198000490665436,
+ "num_input_tokens_seen": 140440,
+ "step": 149,
+ "train_runtime": 539.3165,
+ "train_tokens_per_second": 260.404
+ },
+ {
+ "epoch": 0.03550295857988166,
+ "grad_norm": 0.6818353533744812,
+ "learning_rate": 4.648648648648649e-05,
+ "loss": 0.0710655227303505,
+ "num_input_tokens_seen": 141344,
+ "step": 150,
+ "train_runtime": 542.3824,
+ "train_tokens_per_second": 260.598
+ },
+ {
+ "epoch": 0.0357396449704142,
+ "grad_norm": 0.5292275547981262,
+ "learning_rate": 4.540540540540541e-05,
+ "loss": 0.04743894189596176,
+ "num_input_tokens_seen": 142242,
+ "step": 151,
+ "train_runtime": 549.1326,
+ "train_tokens_per_second": 259.03
+ },
+ {
+ "epoch": 0.035976331360946745,
+ "grad_norm": 0.9436084032058716,
+ "learning_rate": 4.432432432432433e-05,
+ "loss": 0.10009428858757019,
+ "num_input_tokens_seen": 143104,
+ "step": 152,
+ "train_runtime": 552.245,
+ "train_tokens_per_second": 259.131
+ },
+ {
+ "epoch": 0.03621301775147929,
+ "grad_norm": 0.8146379590034485,
+ "learning_rate": 4.324324324324325e-05,
+ "loss": 0.08397847414016724,
+ "num_input_tokens_seen": 144036,
+ "step": 153,
+ "train_runtime": 555.3194,
+ "train_tokens_per_second": 259.375
+ },
+ {
+ "epoch": 0.03644970414201183,
+ "grad_norm": 0.6370963454246521,
+ "learning_rate": 4.2162162162162164e-05,
+ "loss": 0.0647936463356018,
+ "num_input_tokens_seen": 144978,
+ "step": 154,
+ "train_runtime": 558.5158,
+ "train_tokens_per_second": 259.577
+ },
+ {
+ "epoch": 0.036686390532544376,
+ "grad_norm": 0.7876921892166138,
+ "learning_rate": 4.108108108108109e-05,
+ "loss": 0.10044769942760468,
+ "num_input_tokens_seen": 145932,
+ "step": 155,
+ "train_runtime": 561.6626,
+ "train_tokens_per_second": 259.821
+ },
+ {
+ "epoch": 0.036923076923076927,
+ "grad_norm": 0.5711075067520142,
+ "learning_rate": 4e-05,
+ "loss": 0.0628824308514595,
+ "num_input_tokens_seen": 146822,
+ "step": 156,
+ "train_runtime": 564.7686,
+ "train_tokens_per_second": 259.968
+ },
+ {
+ "epoch": 0.03715976331360947,
+ "grad_norm": 0.6890460848808289,
+ "learning_rate": 3.8918918918918926e-05,
+ "loss": 0.0714673101902008,
+ "num_input_tokens_seen": 147746,
+ "step": 157,
+ "train_runtime": 567.7959,
+ "train_tokens_per_second": 260.21
+ },
+ {
+ "epoch": 0.037396449704142014,
+ "grad_norm": 0.6936295032501221,
+ "learning_rate": 3.783783783783784e-05,
+ "loss": 0.05670904368162155,
+ "num_input_tokens_seen": 148648,
+ "step": 158,
+ "train_runtime": 570.8255,
+ "train_tokens_per_second": 260.409
+ },
+ {
+ "epoch": 0.03763313609467456,
+ "grad_norm": 0.6268077492713928,
+ "learning_rate": 3.675675675675676e-05,
+ "loss": 0.04924366623163223,
+ "num_input_tokens_seen": 149622,
+ "step": 159,
+ "train_runtime": 573.9724,
+ "train_tokens_per_second": 260.678
+ },
+ {
+ "epoch": 0.0378698224852071,
+ "grad_norm": 0.7037569284439087,
+ "learning_rate": 3.567567567567568e-05,
+ "loss": 0.06151504069566727,
+ "num_input_tokens_seen": 150586,
+ "step": 160,
+ "train_runtime": 577.103,
+ "train_tokens_per_second": 260.934
+ },
+ {
+ "epoch": 0.038106508875739645,
+ "grad_norm": 0.6471953392028809,
+ "learning_rate": 3.45945945945946e-05,
+ "loss": 0.040921665728092194,
+ "num_input_tokens_seen": 151442,
+ "step": 161,
+ "train_runtime": 580.1219,
+ "train_tokens_per_second": 261.052
+ },
+ {
+ "epoch": 0.03834319526627219,
+ "grad_norm": 0.7046090960502625,
+ "learning_rate": 3.351351351351352e-05,
+ "loss": 0.05204992741346359,
+ "num_input_tokens_seen": 152350,
+ "step": 162,
+ "train_runtime": 583.1537,
+ "train_tokens_per_second": 261.252
+ },
+ {
+ "epoch": 0.03857988165680473,
+ "grad_norm": 1.0495421886444092,
+ "learning_rate": 3.2432432432432436e-05,
+ "loss": 0.1057022288441658,
+ "num_input_tokens_seen": 153278,
+ "step": 163,
+ "train_runtime": 586.3134,
+ "train_tokens_per_second": 261.427
+ },
+ {
+ "epoch": 0.038816568047337276,
+ "grad_norm": 0.6432052850723267,
+ "learning_rate": 3.135135135135135e-05,
+ "loss": 0.05509451404213905,
+ "num_input_tokens_seen": 154178,
+ "step": 164,
+ "train_runtime": 589.3544,
+ "train_tokens_per_second": 261.605
+ },
+ {
+ "epoch": 0.03905325443786982,
+ "grad_norm": 0.6547747254371643,
+ "learning_rate": 3.0270270270270272e-05,
+ "loss": 0.04602127522230148,
+ "num_input_tokens_seen": 155080,
+ "step": 165,
+ "train_runtime": 592.3642,
+ "train_tokens_per_second": 261.798
+ },
+ {
+ "epoch": 0.03928994082840237,
+ "grad_norm": 0.6700029969215393,
+ "learning_rate": 2.918918918918919e-05,
+ "loss": 0.05721529945731163,
+ "num_input_tokens_seen": 156140,
+ "step": 166,
+ "train_runtime": 595.4897,
+ "train_tokens_per_second": 262.204
+ },
+ {
+ "epoch": 0.039526627218934914,
+ "grad_norm": 0.9145491719245911,
+ "learning_rate": 2.810810810810811e-05,
+ "loss": 0.06747015565633774,
+ "num_input_tokens_seen": 157106,
+ "step": 167,
+ "train_runtime": 598.5891,
+ "train_tokens_per_second": 262.461
+ },
+ {
+ "epoch": 0.03976331360946746,
+ "grad_norm": 0.8798087239265442,
+ "learning_rate": 2.702702702702703e-05,
+ "loss": 0.08232270181179047,
+ "num_input_tokens_seen": 158056,
+ "step": 168,
+ "train_runtime": 601.7315,
+ "train_tokens_per_second": 262.669
+ },
+ {
+ "epoch": 0.04,
+ "grad_norm": 1.0618623495101929,
+ "learning_rate": 2.594594594594595e-05,
+ "loss": 0.09569647908210754,
+ "num_input_tokens_seen": 159060,
+ "step": 169,
+ "train_runtime": 604.8838,
+ "train_tokens_per_second": 262.96
+ },
+ {
+ "epoch": 0.040236686390532544,
+ "grad_norm": 0.6945156455039978,
+ "learning_rate": 2.486486486486487e-05,
+ "loss": 0.0597086027264595,
+ "num_input_tokens_seen": 160054,
+ "step": 170,
+ "train_runtime": 608.004,
+ "train_tokens_per_second": 263.245
+ },
+ {
+ "epoch": 0.04047337278106509,
+ "grad_norm": 0.6867018342018127,
+ "learning_rate": 2.3783783783783785e-05,
+ "loss": 0.053483735769987106,
+ "num_input_tokens_seen": 160992,
+ "step": 171,
+ "train_runtime": 611.0992,
+ "train_tokens_per_second": 263.447
+ },
+ {
+ "epoch": 0.04071005917159763,
+ "grad_norm": 0.9504451751708984,
+ "learning_rate": 2.2702702702702705e-05,
+ "loss": 0.07532191276550293,
+ "num_input_tokens_seen": 161910,
+ "step": 172,
+ "train_runtime": 614.2523,
+ "train_tokens_per_second": 263.589
+ },
+ {
+ "epoch": 0.040946745562130175,
+ "grad_norm": 0.6639146208763123,
+ "learning_rate": 2.1621621621621624e-05,
+ "loss": 0.04415765404701233,
+ "num_input_tokens_seen": 162810,
+ "step": 173,
+ "train_runtime": 617.2712,
+ "train_tokens_per_second": 263.758
+ },
+ {
+ "epoch": 0.04118343195266272,
+ "grad_norm": 0.5948973894119263,
+ "learning_rate": 2.0540540540540544e-05,
+ "loss": 0.04589955136179924,
+ "num_input_tokens_seen": 163792,
+ "step": 174,
+ "train_runtime": 620.3367,
+ "train_tokens_per_second": 264.037
+ },
+ {
+ "epoch": 0.04142011834319527,
+ "grad_norm": 0.8319454193115234,
+ "learning_rate": 1.9459459459459463e-05,
+ "loss": 0.07309269905090332,
+ "num_input_tokens_seen": 164770,
+ "step": 175,
+ "train_runtime": 623.4097,
+ "train_tokens_per_second": 264.305
+ },
+ {
+ "epoch": 0.04165680473372781,
+ "grad_norm": 0.591090202331543,
+ "learning_rate": 1.837837837837838e-05,
+ "loss": 0.04063474386930466,
+ "num_input_tokens_seen": 165708,
+ "step": 176,
+ "train_runtime": 626.513,
+ "train_tokens_per_second": 264.493
+ },
+ {
+ "epoch": 0.04189349112426036,
+ "grad_norm": 0.5455971360206604,
+ "learning_rate": 1.72972972972973e-05,
+ "loss": 0.0511426106095314,
+ "num_input_tokens_seen": 166654,
+ "step": 177,
+ "train_runtime": 629.5606,
+ "train_tokens_per_second": 264.715
+ },
+ {
+ "epoch": 0.0421301775147929,
+ "grad_norm": 0.7040805816650391,
+ "learning_rate": 1.6216216216216218e-05,
+ "loss": 0.06684890389442444,
+ "num_input_tokens_seen": 167586,
+ "step": 178,
+ "train_runtime": 632.5911,
+ "train_tokens_per_second": 264.92
+ },
+ {
+ "epoch": 0.042366863905325444,
+ "grad_norm": 0.9840870499610901,
+ "learning_rate": 1.5135135135135136e-05,
+ "loss": 0.08544427156448364,
+ "num_input_tokens_seen": 168558,
+ "step": 179,
+ "train_runtime": 635.6343,
+ "train_tokens_per_second": 265.181
+ },
+ {
+ "epoch": 0.04260355029585799,
+ "grad_norm": 0.763282835483551,
+ "learning_rate": 1.4054054054054055e-05,
+ "loss": 0.05402141064405441,
+ "num_input_tokens_seen": 169594,
+ "step": 180,
+ "train_runtime": 638.8133,
+ "train_tokens_per_second": 265.483
+ },
+ {
+ "epoch": 0.04284023668639053,
+ "grad_norm": 0.5099300742149353,
+ "learning_rate": 1.2972972972972975e-05,
+ "loss": 0.0340673066675663,
+ "num_input_tokens_seen": 170520,
+ "step": 181,
+ "train_runtime": 645.4574,
+ "train_tokens_per_second": 264.185
+ },
+ {
+ "epoch": 0.043076923076923075,
+ "grad_norm": 0.5777792930603027,
+ "learning_rate": 1.1891891891891893e-05,
+ "loss": 0.03730103373527527,
+ "num_input_tokens_seen": 171406,
+ "step": 182,
+ "train_runtime": 648.6265,
+ "train_tokens_per_second": 264.26
+ },
+ {
+ "epoch": 0.04331360946745562,
+ "grad_norm": 0.6748155355453491,
+ "learning_rate": 1.0810810810810812e-05,
+ "loss": 0.04920109361410141,
+ "num_input_tokens_seen": 172352,
+ "step": 183,
+ "train_runtime": 651.7436,
+ "train_tokens_per_second": 264.448
+ },
+ {
+ "epoch": 0.04355029585798816,
+ "grad_norm": 0.7281784415245056,
+ "learning_rate": 9.729729729729732e-06,
+ "loss": 0.06326033174991608,
+ "num_input_tokens_seen": 173274,
+ "step": 184,
+ "train_runtime": 654.8849,
+ "train_tokens_per_second": 264.587
+ },
+ {
+ "epoch": 0.04378698224852071,
+ "grad_norm": 0.4785560667514801,
+ "learning_rate": 8.64864864864865e-06,
+ "loss": 0.03624718636274338,
+ "num_input_tokens_seen": 174218,
+ "step": 185,
+ "train_runtime": 657.9406,
+ "train_tokens_per_second": 264.793
+ },
+ {
+ "epoch": 0.04402366863905326,
+ "grad_norm": 0.6069119572639465,
+ "learning_rate": 7.567567567567568e-06,
+ "loss": 0.055947862565517426,
+ "num_input_tokens_seen": 175232,
+ "step": 186,
+ "train_runtime": 661.0167,
+ "train_tokens_per_second": 265.095
+ },
+ {
+ "epoch": 0.0442603550295858,
+ "grad_norm": 0.4526364207267761,
+ "learning_rate": 6.486486486486487e-06,
+ "loss": 0.03549381345510483,
+ "num_input_tokens_seen": 176176,
+ "step": 187,
+ "train_runtime": 664.2236,
+ "train_tokens_per_second": 265.236
+ },
+ {
+ "epoch": 0.044497041420118344,
+ "grad_norm": 0.5671166777610779,
+ "learning_rate": 5.405405405405406e-06,
+ "loss": 0.04723510146141052,
+ "num_input_tokens_seen": 177148,
+ "step": 188,
+ "train_runtime": 667.3473,
+ "train_tokens_per_second": 265.451
+ },
+ {
+ "epoch": 0.04473372781065089,
+ "grad_norm": 0.4677238464355469,
+ "learning_rate": 4.324324324324325e-06,
+ "loss": 0.03479735553264618,
+ "num_input_tokens_seen": 178120,
+ "step": 189,
+ "train_runtime": 670.3812,
+ "train_tokens_per_second": 265.7
+ },
+ {
+ "epoch": 0.04497041420118343,
+ "grad_norm": 0.7161563038825989,
+ "learning_rate": 3.2432432432432437e-06,
+ "loss": 0.059224192053079605,
+ "num_input_tokens_seen": 178978,
+ "step": 190,
+ "train_runtime": 673.5302,
+ "train_tokens_per_second": 265.731
+ }
+ ],
+ "logging_steps": 1,
+ "max_steps": 190,
+ "num_input_tokens_seen": 178978,
+ "num_train_epochs": 1,
+ "save_steps": 30,
+ "stateful_callbacks": {
+ "TrainerControl": {
+ "args": {
+ "should_epoch_stop": false,
+ "should_evaluate": false,
+ "should_log": false,
+ "should_save": true,
+ "should_training_stop": true
+ },
+ "attributes": {}
+ }
+ },
+ "total_flos": 2580240718381440.0,
+ "train_batch_size": 2,
+ "trial_name": null,
+ "trial_params": null
+}
diff --git a/checkpoint-190/training_args.bin b/checkpoint-190/training_args.bin
new file mode 100644
index 0000000000000000000000000000000000000000..cefffe2bb567bdf18003e5cc5aaf746472c2981f
--- /dev/null
+++ b/checkpoint-190/training_args.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:34c4ca6d50f04301fb38588516a91421e14a8ad1ffddf12afdfd9ba35cd4b64e
+size 5777
diff --git a/checkpoint-30/README.md b/checkpoint-30/README.md
new file mode 100644
index 0000000000000000000000000000000000000000..41470ec6557bbc37155893f70bd59986401b9148
--- /dev/null
+++ b/checkpoint-30/README.md
@@ -0,0 +1,210 @@
+---
+base_model: unsloth/gemma-4-e2b-it-unsloth-bnb-4bit
+library_name: peft
+pipeline_tag: text-generation
+tags:
+- base_model:adapter:unsloth/gemma-4-e2b-it-unsloth-bnb-4bit
+- lora
+- sft
+- transformers
+- trl
+- unsloth
+---
+
+# Model Card for Model ID
+
+
+
+
+
+## Model Details
+
+### Model Description
+
+
+
+
+
+- **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]
+
+
+
+- **Repository:** [More Information Needed]
+- **Paper [optional]:** [More Information Needed]
+- **Demo [optional]:** [More Information Needed]
+
+## Uses
+
+
+
+### Direct Use
+
+
+
+[More Information Needed]
+
+### Downstream Use [optional]
+
+
+
+[More Information Needed]
+
+### Out-of-Scope Use
+
+
+
+[More Information Needed]
+
+## Bias, Risks, and Limitations
+
+
+
+[More Information Needed]
+
+### Recommendations
+
+
+
+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
+
+
+
+[More Information Needed]
+
+### Training Procedure
+
+
+
+#### Preprocessing [optional]
+
+[More Information Needed]
+
+
+#### Training Hyperparameters
+
+- **Training regime:** [More Information Needed]
+
+#### Speeds, Sizes, Times [optional]
+
+
+
+[More Information Needed]
+
+## Evaluation
+
+
+
+### Testing Data, Factors & Metrics
+
+#### Testing Data
+
+
+
+[More Information Needed]
+
+#### Factors
+
+
+
+[More Information Needed]
+
+#### Metrics
+
+
+
+[More Information Needed]
+
+### Results
+
+[More Information Needed]
+
+#### Summary
+
+
+
+## Model Examination [optional]
+
+
+
+[More Information Needed]
+
+## Environmental Impact
+
+
+
+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]
+
+
+
+**BibTeX:**
+
+[More Information Needed]
+
+**APA:**
+
+[More Information Needed]
+
+## Glossary [optional]
+
+
+
+[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
\ No newline at end of file
diff --git a/checkpoint-30/adapter_config.json b/checkpoint-30/adapter_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..915726be1cd67eb165724a8390b51a57fe0dd6df
--- /dev/null
+++ b/checkpoint-30/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": "unsloth/gemma-4-e2b-it-unsloth-bnb-4bit",
+ "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": [
+ "q_proj",
+ "gate_proj",
+ "v_proj",
+ "k_proj",
+ "down_proj",
+ "o_proj",
+ "up_proj"
+ ],
+ "target_parameters": null,
+ "task_type": "CAUSAL_LM",
+ "trainable_token_indices": null,
+ "use_dora": false,
+ "use_qalora": false,
+ "use_rslora": false
+}
\ No newline at end of file
diff --git a/checkpoint-30/adapter_model.safetensors b/checkpoint-30/adapter_model.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..f371492d1e4bb27092656efb7025db1e9b3785cc
--- /dev/null
+++ b/checkpoint-30/adapter_model.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:15498a59af388d964937099160eda9f496174a1b9ac196180b8df7b32e165a7a
+size 124277728
diff --git a/checkpoint-30/chat_template.jinja b/checkpoint-30/chat_template.jinja
new file mode 100644
index 0000000000000000000000000000000000000000..dc032e46e28e0d325b18d26da1279cdda3afa8d7
--- /dev/null
+++ b/checkpoint-30/chat_template.jinja
@@ -0,0 +1,351 @@
+{%- macro format_parameters(properties, required, filter_keys=false) -%}
+ {%- 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 not filter_keys or 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([]), filter_keys=true) -}}
+ }
+ {%- 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('') -%}
+ {%- 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 -%}
+ {{- '' -}}
+{%- 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'] -%}
+ {%- if messages[0]['content'] is string -%}
+ {{- messages[0]['content'] | trim -}}
+ {%- elif messages[0]['content'] is sequence -%}
+ {%- for item in messages[0]['content'] -%}
+ {{- item['text'] | trim + ' '-}}
+ {%- endfor -%}
+ {%- endif -%}
+ {%- set loop_messages = messages[1:] -%}
+ {%- endif -%}
+ {%- if tools -%}
+ {%- for tool in tools %}
+ {{- '<|tool>' -}}
+ {{- format_function_declaration(tool) | trim -}}
+ {{- '' -}}
+ {%- endfor %}
+ {%- set ns.prev_message_type = 'tool' -%}
+ {%- endif -%}
+ {{- '\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' -}}
+ {%- 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 -%}
+ {{- '}' -}}
+ {%- 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 -%}
+
+ {%- set captured_content -%}
+ {%- 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 -%}
+ {%- endset -%}
+
+ {{- captured_content -}}
+ {%- set has_content = captured_content | trim | length > 0 -%}
+
+ {%- if ns.prev_message_type == 'tool_call' and not ns_tr_out.flag -%}
+ {{- '<|tool_response>' -}}
+ {%- elif not (ns_tr_out.flag and not has_content) -%}
+ {{- '\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 -%}
\ No newline at end of file
diff --git a/checkpoint-30/optimizer.pt b/checkpoint-30/optimizer.pt
new file mode 100644
index 0000000000000000000000000000000000000000..e299f492118fa4ab1fe1d96f38fc44134cd43967
--- /dev/null
+++ b/checkpoint-30/optimizer.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:cd7373b9baf1b47dac0846b30e03923d6194d2d63b39ea1c451c4a21d5987014
+size 52047117
diff --git a/checkpoint-30/processor_config.json b/checkpoint-30/processor_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..bfaa5e360ad17c2761b7ee3b05abf03e381e8aa1
--- /dev/null
+++ b/checkpoint-30/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"
+ }
+}
diff --git a/checkpoint-30/rng_state.pth b/checkpoint-30/rng_state.pth
new file mode 100644
index 0000000000000000000000000000000000000000..3ef66339b9befa098183fd5d69faed6838e526b0
--- /dev/null
+++ b/checkpoint-30/rng_state.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f1d565802a8e26c4e8a31328752b7a7fdc186d9401aa008e65697d0ad8c22e33
+size 14645
diff --git a/checkpoint-30/scaler.pt b/checkpoint-30/scaler.pt
new file mode 100644
index 0000000000000000000000000000000000000000..802cc24dfe5217a48ca4920be1e5978d3d11826d
--- /dev/null
+++ b/checkpoint-30/scaler.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b5ad10c944e2d21aeef682a674c8074069dd68f583dd60c49b218e2d2739e279
+size 1383
diff --git a/checkpoint-30/scheduler.pt b/checkpoint-30/scheduler.pt
new file mode 100644
index 0000000000000000000000000000000000000000..2c4e379878c5e773093a6b3b0449975ffea004f1
--- /dev/null
+++ b/checkpoint-30/scheduler.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:88b2e7f1ce36b7233925b64bc26e328a32a0215bfd0b00275a3a317a8271f33a
+size 1465
diff --git a/checkpoint-30/tokenizer.json b/checkpoint-30/tokenizer.json
new file mode 100644
index 0000000000000000000000000000000000000000..1ff9f3e3439a939b971f9919e821bf87e835a503
--- /dev/null
+++ b/checkpoint-30/tokenizer.json
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:cc8d3a0ce36466ccc1278bf987df5f71db1719b9ca6b4118264f45cb627bfe0f
+size 32169626
diff --git a/checkpoint-30/tokenizer_config.json b/checkpoint-30/tokenizer_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..24b1522bb32b0eb05b631b1a9f112751b8cb0fab
--- /dev/null
+++ b/checkpoint-30/tokenizer_config.json
@@ -0,0 +1,289 @@
+{
+ "audio_token": "<|audio|>",
+ "backend": "tokenizers",
+ "boa_token": "<|audio>",
+ "boi_token": "<|image>",
+ "bos_token": "",
+ "eoa_token": "",
+ "eoc_token": "",
+ "eoi_token": "",
+ "eos_token": "",
+ "eot_token": "",
+ "escape_token": "<|\"|>",
+ "etc_token": "",
+ "etd_token": "",
+ "etr_token": "",
+ "extra_special_tokens": [
+ "<|video|>"
+ ],
+ "image_token": "<|image|>",
+ "is_local": false,
+ "mask_token": "",
+ "model_max_length": 131072,
+ "model_specific_special_tokens": {
+ "audio_token": "<|audio|>",
+ "boa_token": "<|audio>",
+ "boi_token": "<|image>",
+ "eoa_token": "",
+ "eoc_token": "",
+ "eoi_token": "",
+ "eot_token": "",
+ "escape_token": "<|\"|>",
+ "etc_token": "",
+ "etd_token": "",
+ "etr_token": "",
+ "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": "",
+ "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\\w+)(?P\\{.*\\})"
+ },
+ "type": {
+ "const": "function"
+ }
+ },
+ "type": "object"
+ },
+ "type": "array",
+ "x-regex-iterator": "<\\|tool_call>(.*?)"
+ }
+ },
+ "type": "object",
+ "x-regex": "(\\<\\|channel\\>thought\\n(?P.*?)\\)?(?P\\<\\|tool_call\\>.*\\)?(?P(?:(?!\\)(?!\\<\\|tool_response\\>).)+)?(?:\\|\\<\\|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": "",
+ "added_tokens_decoder": {
+ "0": {
+ "content": "",
+ "single_word": false,
+ "lstrip": false,
+ "rstrip": false,
+ "normalized": false,
+ "special": true
+ },
+ "1": {
+ "content": "",
+ "single_word": false,
+ "lstrip": false,
+ "rstrip": false,
+ "normalized": false,
+ "special": true
+ },
+ "2": {
+ "content": "",
+ "single_word": false,
+ "lstrip": false,
+ "rstrip": false,
+ "normalized": false,
+ "special": true
+ },
+ "3": {
+ "content": "",
+ "single_word": false,
+ "lstrip": false,
+ "rstrip": false,
+ "normalized": false,
+ "special": true
+ },
+ "4": {
+ "content": "",
+ "single_word": false,
+ "lstrip": false,
+ "rstrip": false,
+ "normalized": false,
+ "special": true
+ },
+ "46": {
+ "content": "<|tool>",
+ "single_word": false,
+ "lstrip": false,
+ "rstrip": false,
+ "normalized": false,
+ "special": true
+ },
+ "47": {
+ "content": "",
+ "single_word": false,
+ "lstrip": false,
+ "rstrip": false,
+ "normalized": false,
+ "special": true
+ },
+ "48": {
+ "content": "<|tool_call>",
+ "single_word": false,
+ "lstrip": false,
+ "rstrip": false,
+ "normalized": false,
+ "special": true
+ },
+ "49": {
+ "content": "",
+ "single_word": false,
+ "lstrip": false,
+ "rstrip": false,
+ "normalized": false,
+ "special": true
+ },
+ "50": {
+ "content": "<|tool_response>",
+ "single_word": false,
+ "lstrip": false,
+ "rstrip": false,
+ "normalized": false,
+ "special": true
+ },
+ "51": {
+ "content": "",
+ "single_word": false,
+ "lstrip": false,
+ "rstrip": false,
+ "normalized": false,
+ "special": true
+ },
+ "52": {
+ "content": "<|\"|>",
+ "single_word": false,
+ "lstrip": false,
+ "rstrip": false,
+ "normalized": false,
+ "special": true
+ },
+ "98": {
+ "content": "<|think|>",
+ "single_word": false,
+ "lstrip": false,
+ "rstrip": false,
+ "normalized": false,
+ "special": true
+ },
+ "100": {
+ "content": "<|channel>",
+ "single_word": false,
+ "lstrip": false,
+ "rstrip": false,
+ "normalized": false,
+ "special": true
+ },
+ "101": {
+ "content": "",
+ "single_word": false,
+ "lstrip": false,
+ "rstrip": false,
+ "normalized": false,
+ "special": true
+ },
+ "105": {
+ "content": "<|turn>",
+ "single_word": false,
+ "lstrip": false,
+ "rstrip": false,
+ "normalized": false,
+ "special": true
+ },
+ "106": {
+ "content": "",
+ "single_word": false,
+ "lstrip": false,
+ "rstrip": false,
+ "normalized": false,
+ "special": true
+ },
+ "255999": {
+ "content": "<|image>",
+ "single_word": false,
+ "lstrip": false,
+ "rstrip": false,
+ "normalized": false,
+ "special": true
+ },
+ "256000": {
+ "content": "<|audio>",
+ "single_word": false,
+ "lstrip": false,
+ "rstrip": false,
+ "normalized": false,
+ "special": true
+ },
+ "258880": {
+ "content": "<|image|>",
+ "single_word": false,
+ "lstrip": false,
+ "rstrip": false,
+ "normalized": false,
+ "special": true
+ },
+ "258881": {
+ "content": "<|audio|>",
+ "single_word": false,
+ "lstrip": false,
+ "rstrip": false,
+ "normalized": false,
+ "special": true
+ },
+ "258882": {
+ "content": "",
+ "single_word": false,
+ "lstrip": false,
+ "rstrip": false,
+ "normalized": false,
+ "special": true
+ },
+ "258883": {
+ "content": "",
+ "single_word": false,
+ "lstrip": false,
+ "rstrip": false,
+ "normalized": false,
+ "special": true
+ },
+ "258884": {
+ "content": "<|video|>",
+ "single_word": false,
+ "lstrip": false,
+ "rstrip": false,
+ "normalized": false,
+ "special": true
+ }
+ }
+}
diff --git a/checkpoint-30/trainer_state.json b/checkpoint-30/trainer_state.json
new file mode 100644
index 0000000000000000000000000000000000000000..c392bc15b0d591d7aa1f8794e3f8a8386cdb334d
--- /dev/null
+++ b/checkpoint-30/trainer_state.json
@@ -0,0 +1,334 @@
+{
+ "best_global_step": null,
+ "best_metric": null,
+ "best_model_checkpoint": null,
+ "epoch": 0.007100591715976331,
+ "eval_steps": 500,
+ "global_step": 30,
+ "is_hyper_param_search": false,
+ "is_local_process_zero": true,
+ "is_world_process_zero": true,
+ "log_history": [
+ {
+ "epoch": 0.00023668639053254438,
+ "grad_norm": NaN,
+ "learning_rate": 0.0,
+ "loss": 4.347962379455566,
+ "num_input_tokens_seen": 842,
+ "step": 1,
+ "train_runtime": 63.0289,
+ "train_tokens_per_second": 13.359
+ },
+ {
+ "epoch": 0.00047337278106508875,
+ "grad_norm": NaN,
+ "learning_rate": 0.0,
+ "loss": 4.634688854217529,
+ "num_input_tokens_seen": 1820,
+ "step": 2,
+ "train_runtime": 67.7033,
+ "train_tokens_per_second": 26.882
+ },
+ {
+ "epoch": 0.0007100591715976331,
+ "grad_norm": 13.944212913513184,
+ "learning_rate": 0.0,
+ "loss": 4.42241096496582,
+ "num_input_tokens_seen": 2754,
+ "step": 3,
+ "train_runtime": 72.324,
+ "train_tokens_per_second": 38.079
+ },
+ {
+ "epoch": 0.0009467455621301775,
+ "grad_norm": 13.445413589477539,
+ "learning_rate": 4e-05,
+ "loss": 4.682804107666016,
+ "num_input_tokens_seen": 3700,
+ "step": 4,
+ "train_runtime": 75.3925,
+ "train_tokens_per_second": 49.077
+ },
+ {
+ "epoch": 0.001183431952662722,
+ "grad_norm": 10.576727867126465,
+ "learning_rate": 8e-05,
+ "loss": 4.166159629821777,
+ "num_input_tokens_seen": 4714,
+ "step": 5,
+ "train_runtime": 78.474,
+ "train_tokens_per_second": 60.071
+ },
+ {
+ "epoch": 0.0014201183431952662,
+ "grad_norm": 8.077850341796875,
+ "learning_rate": 0.00012,
+ "loss": 3.2006261348724365,
+ "num_input_tokens_seen": 5674,
+ "step": 6,
+ "train_runtime": 81.6139,
+ "train_tokens_per_second": 69.522
+ },
+ {
+ "epoch": 0.0016568047337278107,
+ "grad_norm": 6.60789680480957,
+ "learning_rate": 0.00016,
+ "loss": 2.7067415714263916,
+ "num_input_tokens_seen": 6646,
+ "step": 7,
+ "train_runtime": 84.786,
+ "train_tokens_per_second": 78.386
+ },
+ {
+ "epoch": 0.001893491124260355,
+ "grad_norm": 9.175209045410156,
+ "learning_rate": 0.0002,
+ "loss": 2.367570638656616,
+ "num_input_tokens_seen": 7518,
+ "step": 8,
+ "train_runtime": 87.8178,
+ "train_tokens_per_second": 85.609
+ },
+ {
+ "epoch": 0.0021301775147928993,
+ "grad_norm": 3.438948631286621,
+ "learning_rate": 0.00019891891891891895,
+ "loss": 2.1013119220733643,
+ "num_input_tokens_seen": 8470,
+ "step": 9,
+ "train_runtime": 90.9458,
+ "train_tokens_per_second": 93.132
+ },
+ {
+ "epoch": 0.002366863905325444,
+ "grad_norm": 3.459472894668579,
+ "learning_rate": 0.00019783783783783784,
+ "loss": 1.8991940021514893,
+ "num_input_tokens_seen": 9458,
+ "step": 10,
+ "train_runtime": 94.0577,
+ "train_tokens_per_second": 100.555
+ },
+ {
+ "epoch": 0.0026035502958579883,
+ "grad_norm": 2.169750928878784,
+ "learning_rate": 0.00019675675675675678,
+ "loss": 1.7395055294036865,
+ "num_input_tokens_seen": 10456,
+ "step": 11,
+ "train_runtime": 97.2068,
+ "train_tokens_per_second": 107.565
+ },
+ {
+ "epoch": 0.0028402366863905324,
+ "grad_norm": 1.8638927936553955,
+ "learning_rate": 0.00019567567567567567,
+ "loss": 1.5622950792312622,
+ "num_input_tokens_seen": 11410,
+ "step": 12,
+ "train_runtime": 100.2911,
+ "train_tokens_per_second": 113.769
+ },
+ {
+ "epoch": 0.003076923076923077,
+ "grad_norm": 1.4741116762161255,
+ "learning_rate": 0.00019459459459459462,
+ "loss": 1.392699956893921,
+ "num_input_tokens_seen": 12316,
+ "step": 13,
+ "train_runtime": 103.3664,
+ "train_tokens_per_second": 119.149
+ },
+ {
+ "epoch": 0.0033136094674556214,
+ "grad_norm": 1.3238699436187744,
+ "learning_rate": 0.00019351351351351353,
+ "loss": 1.2535074949264526,
+ "num_input_tokens_seen": 13256,
+ "step": 14,
+ "train_runtime": 106.4031,
+ "train_tokens_per_second": 124.583
+ },
+ {
+ "epoch": 0.0035502958579881655,
+ "grad_norm": 1.276484489440918,
+ "learning_rate": 0.00019243243243243245,
+ "loss": 1.135048508644104,
+ "num_input_tokens_seen": 14228,
+ "step": 15,
+ "train_runtime": 109.4909,
+ "train_tokens_per_second": 129.947
+ },
+ {
+ "epoch": 0.00378698224852071,
+ "grad_norm": 1.1560865640640259,
+ "learning_rate": 0.00019135135135135137,
+ "loss": 1.0317097902297974,
+ "num_input_tokens_seen": 15192,
+ "step": 16,
+ "train_runtime": 112.5741,
+ "train_tokens_per_second": 134.951
+ },
+ {
+ "epoch": 0.004023668639053254,
+ "grad_norm": 1.1331353187561035,
+ "learning_rate": 0.00019027027027027028,
+ "loss": 0.9967219233512878,
+ "num_input_tokens_seen": 16130,
+ "step": 17,
+ "train_runtime": 115.6495,
+ "train_tokens_per_second": 139.473
+ },
+ {
+ "epoch": 0.004260355029585799,
+ "grad_norm": 1.1567702293395996,
+ "learning_rate": 0.0001891891891891892,
+ "loss": 0.905368447303772,
+ "num_input_tokens_seen": 17078,
+ "step": 18,
+ "train_runtime": 118.6796,
+ "train_tokens_per_second": 143.9
+ },
+ {
+ "epoch": 0.004497041420118343,
+ "grad_norm": 0.9157600998878479,
+ "learning_rate": 0.00018810810810810814,
+ "loss": 0.8075690865516663,
+ "num_input_tokens_seen": 18062,
+ "step": 19,
+ "train_runtime": 121.8602,
+ "train_tokens_per_second": 148.219
+ },
+ {
+ "epoch": 0.004733727810650888,
+ "grad_norm": 0.8306605815887451,
+ "learning_rate": 0.00018702702702702703,
+ "loss": 0.8007175922393799,
+ "num_input_tokens_seen": 19078,
+ "step": 20,
+ "train_runtime": 125.0019,
+ "train_tokens_per_second": 152.622
+ },
+ {
+ "epoch": 0.004970414201183432,
+ "grad_norm": 0.690018355846405,
+ "learning_rate": 0.00018594594594594597,
+ "loss": 0.7810791730880737,
+ "num_input_tokens_seen": 20068,
+ "step": 21,
+ "train_runtime": 128.0932,
+ "train_tokens_per_second": 156.667
+ },
+ {
+ "epoch": 0.005207100591715977,
+ "grad_norm": 1.4768867492675781,
+ "learning_rate": 0.00018486486486486486,
+ "loss": 0.7172120809555054,
+ "num_input_tokens_seen": 20972,
+ "step": 22,
+ "train_runtime": 131.0901,
+ "train_tokens_per_second": 159.982
+ },
+ {
+ "epoch": 0.00544378698224852,
+ "grad_norm": 0.7999809980392456,
+ "learning_rate": 0.0001837837837837838,
+ "loss": 0.6333726048469543,
+ "num_input_tokens_seen": 21900,
+ "step": 23,
+ "train_runtime": 134.212,
+ "train_tokens_per_second": 163.175
+ },
+ {
+ "epoch": 0.005680473372781065,
+ "grad_norm": 0.9285996556282043,
+ "learning_rate": 0.00018270270270270272,
+ "loss": 0.6750729084014893,
+ "num_input_tokens_seen": 22770,
+ "step": 24,
+ "train_runtime": 137.3193,
+ "train_tokens_per_second": 165.818
+ },
+ {
+ "epoch": 0.005917159763313609,
+ "grad_norm": 0.7733820080757141,
+ "learning_rate": 0.00018162162162162164,
+ "loss": 0.672032356262207,
+ "num_input_tokens_seen": 23746,
+ "step": 25,
+ "train_runtime": 140.3175,
+ "train_tokens_per_second": 169.23
+ },
+ {
+ "epoch": 0.006153846153846154,
+ "grad_norm": 0.73944091796875,
+ "learning_rate": 0.00018054054054054055,
+ "loss": 0.6716101169586182,
+ "num_input_tokens_seen": 24716,
+ "step": 26,
+ "train_runtime": 143.3683,
+ "train_tokens_per_second": 172.395
+ },
+ {
+ "epoch": 0.006390532544378698,
+ "grad_norm": 0.7659372091293335,
+ "learning_rate": 0.00017945945945945947,
+ "loss": 0.6067175269126892,
+ "num_input_tokens_seen": 25724,
+ "step": 27,
+ "train_runtime": 146.5377,
+ "train_tokens_per_second": 175.545
+ },
+ {
+ "epoch": 0.006627218934911243,
+ "grad_norm": 0.5428617000579834,
+ "learning_rate": 0.00017837837837837839,
+ "loss": 0.6083395481109619,
+ "num_input_tokens_seen": 26712,
+ "step": 28,
+ "train_runtime": 149.6429,
+ "train_tokens_per_second": 178.505
+ },
+ {
+ "epoch": 0.006863905325443787,
+ "grad_norm": 0.5915114283561707,
+ "learning_rate": 0.00017729729729729733,
+ "loss": 0.6463742256164551,
+ "num_input_tokens_seen": 27704,
+ "step": 29,
+ "train_runtime": 152.7501,
+ "train_tokens_per_second": 181.368
+ },
+ {
+ "epoch": 0.007100591715976331,
+ "grad_norm": 1.3350757360458374,
+ "learning_rate": 0.00017621621621621622,
+ "loss": 0.5705698132514954,
+ "num_input_tokens_seen": 28648,
+ "step": 30,
+ "train_runtime": 155.8918,
+ "train_tokens_per_second": 183.768
+ }
+ ],
+ "logging_steps": 1,
+ "max_steps": 190,
+ "num_input_tokens_seen": 28648,
+ "num_train_epochs": 1,
+ "save_steps": 30,
+ "stateful_callbacks": {
+ "TrainerControl": {
+ "args": {
+ "should_epoch_stop": false,
+ "should_evaluate": false,
+ "should_log": false,
+ "should_save": true,
+ "should_training_stop": false
+ },
+ "attributes": {}
+ }
+ },
+ "total_flos": 413004593303040.0,
+ "train_batch_size": 2,
+ "trial_name": null,
+ "trial_params": null
+}
diff --git a/checkpoint-30/training_args.bin b/checkpoint-30/training_args.bin
new file mode 100644
index 0000000000000000000000000000000000000000..cefffe2bb567bdf18003e5cc5aaf746472c2981f
--- /dev/null
+++ b/checkpoint-30/training_args.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:34c4ca6d50f04301fb38588516a91421e14a8ad1ffddf12afdfd9ba35cd4b64e
+size 5777
diff --git a/checkpoint-60/README.md b/checkpoint-60/README.md
new file mode 100644
index 0000000000000000000000000000000000000000..41470ec6557bbc37155893f70bd59986401b9148
--- /dev/null
+++ b/checkpoint-60/README.md
@@ -0,0 +1,210 @@
+---
+base_model: unsloth/gemma-4-e2b-it-unsloth-bnb-4bit
+library_name: peft
+pipeline_tag: text-generation
+tags:
+- base_model:adapter:unsloth/gemma-4-e2b-it-unsloth-bnb-4bit
+- lora
+- sft
+- transformers
+- trl
+- unsloth
+---
+
+# Model Card for Model ID
+
+
+
+
+
+## Model Details
+
+### Model Description
+
+
+
+
+
+- **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]
+
+
+
+- **Repository:** [More Information Needed]
+- **Paper [optional]:** [More Information Needed]
+- **Demo [optional]:** [More Information Needed]
+
+## Uses
+
+
+
+### Direct Use
+
+
+
+[More Information Needed]
+
+### Downstream Use [optional]
+
+
+
+[More Information Needed]
+
+### Out-of-Scope Use
+
+
+
+[More Information Needed]
+
+## Bias, Risks, and Limitations
+
+
+
+[More Information Needed]
+
+### Recommendations
+
+
+
+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
+
+
+
+[More Information Needed]
+
+### Training Procedure
+
+
+
+#### Preprocessing [optional]
+
+[More Information Needed]
+
+
+#### Training Hyperparameters
+
+- **Training regime:** [More Information Needed]
+
+#### Speeds, Sizes, Times [optional]
+
+
+
+[More Information Needed]
+
+## Evaluation
+
+
+
+### Testing Data, Factors & Metrics
+
+#### Testing Data
+
+
+
+[More Information Needed]
+
+#### Factors
+
+
+
+[More Information Needed]
+
+#### Metrics
+
+
+
+[More Information Needed]
+
+### Results
+
+[More Information Needed]
+
+#### Summary
+
+
+
+## Model Examination [optional]
+
+
+
+[More Information Needed]
+
+## Environmental Impact
+
+
+
+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]
+
+
+
+**BibTeX:**
+
+[More Information Needed]
+
+**APA:**
+
+[More Information Needed]
+
+## Glossary [optional]
+
+
+
+[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
\ No newline at end of file
diff --git a/checkpoint-60/adapter_config.json b/checkpoint-60/adapter_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..915726be1cd67eb165724a8390b51a57fe0dd6df
--- /dev/null
+++ b/checkpoint-60/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": "unsloth/gemma-4-e2b-it-unsloth-bnb-4bit",
+ "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": [
+ "q_proj",
+ "gate_proj",
+ "v_proj",
+ "k_proj",
+ "down_proj",
+ "o_proj",
+ "up_proj"
+ ],
+ "target_parameters": null,
+ "task_type": "CAUSAL_LM",
+ "trainable_token_indices": null,
+ "use_dora": false,
+ "use_qalora": false,
+ "use_rslora": false
+}
\ No newline at end of file
diff --git a/checkpoint-60/adapter_model.safetensors b/checkpoint-60/adapter_model.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..d4e5443c8a5a7762b016a27427fb41740d624454
--- /dev/null
+++ b/checkpoint-60/adapter_model.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:798e54e773414574465af4ec9decee158fd221f78903a7e0823dd3760e506d13
+size 124277728
diff --git a/checkpoint-60/chat_template.jinja b/checkpoint-60/chat_template.jinja
new file mode 100644
index 0000000000000000000000000000000000000000..dc032e46e28e0d325b18d26da1279cdda3afa8d7
--- /dev/null
+++ b/checkpoint-60/chat_template.jinja
@@ -0,0 +1,351 @@
+{%- macro format_parameters(properties, required, filter_keys=false) -%}
+ {%- 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 not filter_keys or 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([]), filter_keys=true) -}}
+ }
+ {%- 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('') -%}
+ {%- 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 -%}
+ {{- '' -}}
+{%- 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'] -%}
+ {%- if messages[0]['content'] is string -%}
+ {{- messages[0]['content'] | trim -}}
+ {%- elif messages[0]['content'] is sequence -%}
+ {%- for item in messages[0]['content'] -%}
+ {{- item['text'] | trim + ' '-}}
+ {%- endfor -%}
+ {%- endif -%}
+ {%- set loop_messages = messages[1:] -%}
+ {%- endif -%}
+ {%- if tools -%}
+ {%- for tool in tools %}
+ {{- '<|tool>' -}}
+ {{- format_function_declaration(tool) | trim -}}
+ {{- '' -}}
+ {%- endfor %}
+ {%- set ns.prev_message_type = 'tool' -%}
+ {%- endif -%}
+ {{- '\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' -}}
+ {%- 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 -%}
+ {{- '}' -}}
+ {%- 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 -%}
+
+ {%- set captured_content -%}
+ {%- 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 -%}
+ {%- endset -%}
+
+ {{- captured_content -}}
+ {%- set has_content = captured_content | trim | length > 0 -%}
+
+ {%- if ns.prev_message_type == 'tool_call' and not ns_tr_out.flag -%}
+ {{- '<|tool_response>' -}}
+ {%- elif not (ns_tr_out.flag and not has_content) -%}
+ {{- '\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 -%}
\ No newline at end of file
diff --git a/checkpoint-60/optimizer.pt b/checkpoint-60/optimizer.pt
new file mode 100644
index 0000000000000000000000000000000000000000..1eae1f319b6c88abc7e7780c519edbc048068b8e
--- /dev/null
+++ b/checkpoint-60/optimizer.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3c8cc4ebb62e0781821c4a90b7ee14fce935dbec3f5b616d7fce7eedceeb1d15
+size 52047117
diff --git a/checkpoint-60/processor_config.json b/checkpoint-60/processor_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..bfaa5e360ad17c2761b7ee3b05abf03e381e8aa1
--- /dev/null
+++ b/checkpoint-60/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"
+ }
+}
diff --git a/checkpoint-60/rng_state.pth b/checkpoint-60/rng_state.pth
new file mode 100644
index 0000000000000000000000000000000000000000..3ef66339b9befa098183fd5d69faed6838e526b0
--- /dev/null
+++ b/checkpoint-60/rng_state.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f1d565802a8e26c4e8a31328752b7a7fdc186d9401aa008e65697d0ad8c22e33
+size 14645
diff --git a/checkpoint-60/scaler.pt b/checkpoint-60/scaler.pt
new file mode 100644
index 0000000000000000000000000000000000000000..a5c211f5d67fa4959bfe47dd974c369d96816c88
--- /dev/null
+++ b/checkpoint-60/scaler.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:cc0984e75370fd0f920a2690fbfe37926629fa7a6666636f32ed583f054361ef
+size 1383
diff --git a/checkpoint-60/scheduler.pt b/checkpoint-60/scheduler.pt
new file mode 100644
index 0000000000000000000000000000000000000000..f11fd48624699260510e99e0ed10cf719af88708
--- /dev/null
+++ b/checkpoint-60/scheduler.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:333dd3fe3ceed79360c29b6ce11cbcad10f04d68077d2234f8914024308a70ca
+size 1465
diff --git a/checkpoint-60/tokenizer.json b/checkpoint-60/tokenizer.json
new file mode 100644
index 0000000000000000000000000000000000000000..1ff9f3e3439a939b971f9919e821bf87e835a503
--- /dev/null
+++ b/checkpoint-60/tokenizer.json
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:cc8d3a0ce36466ccc1278bf987df5f71db1719b9ca6b4118264f45cb627bfe0f
+size 32169626
diff --git a/checkpoint-60/tokenizer_config.json b/checkpoint-60/tokenizer_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..24b1522bb32b0eb05b631b1a9f112751b8cb0fab
--- /dev/null
+++ b/checkpoint-60/tokenizer_config.json
@@ -0,0 +1,289 @@
+{
+ "audio_token": "<|audio|>",
+ "backend": "tokenizers",
+ "boa_token": "<|audio>",
+ "boi_token": "<|image>",
+ "bos_token": "",
+ "eoa_token": "",
+ "eoc_token": "",
+ "eoi_token": "",
+ "eos_token": "",
+ "eot_token": "",
+ "escape_token": "<|\"|>",
+ "etc_token": "