Add files using upload-large-folder tool

overgeneralisation_original_Estonian/Qwen3.5-4B-Base_overgeneralisation_splits_original_features_train_overgeneralisation_splits_original_features_test1/README.md

@@ -0,0 +1,58 @@
+---
+base_model: Qwen/Qwen3.5-4B-Base
+library_name: transformers
+model_name: Qwen3.5-4B-Base_overgeneralisation_splits_original_features_train_overgeneralisation_splits_original_features_test1
+tags:
+- generated_from_trainer
+- trl
+- sft
+licence: license
+---
+
+# Model Card for Qwen3.5-4B-Base_overgeneralisation_splits_original_features_train_overgeneralisation_splits_original_features_test1
+
+This model is a fine-tuned version of [Qwen/Qwen3.5-4B-Base](https://huggingface.co/Qwen/Qwen3.5-4B-Base).
+It has been trained using [TRL](https://github.com/huggingface/trl).
+
+## Quick start
+
+```python
+from transformers import pipeline
+
+question = "If you had a time machine, but could only go to the past or the future once and never return, which would you choose and why?"
+generator = pipeline("text-generation", model="None", device="cuda")
+output = generator([{"role": "user", "content": question}], max_new_tokens=128, return_full_text=False)[0]
+print(output["generated_text"])
+```
+
+## Training procedure
+
+[<img src="https://raw.githubusercontent.com/wandb/assets/main/wandb-github-badge-28.svg" alt="Visualize in Weights & Biases" width="150" height="24"/>](https://wandb.ai/katriin-kukk/Cross_lingual_morphological_generalization/runs/mq7in2f1) 
+
+
+
+This model was trained with SFT.
+
+### Framework versions
+
+- TRL: 0.29.0
+- Transformers: 5.5.4
+- Pytorch: 2.10.0
+- Datasets: 4.6.1
+- Tokenizers: 0.22.2
+
+## Citations
+
+
+
+Cite TRL as:
+    
+```bibtex
+@software{vonwerra2020trl,
+  title   = {{TRL: Transformers Reinforcement Learning}},
+  author  = {von Werra, Leandro and Belkada, Younes and Tunstall, Lewis and Beeching, Edward and Thrush, Tristan and Lambert, Nathan and Huang, Shengyi and Rasul, Kashif and Gallouédec, Quentin},
+  license = {Apache-2.0},
+  url     = {https://github.com/huggingface/trl},
+  year    = {2020}
+}
+```

overgeneralisation_original_Estonian/Qwen3.5-4B-Base_overgeneralisation_splits_original_features_train_overgeneralisation_splits_original_features_test2/README.md

@@ -0,0 +1,58 @@
+---
+base_model: Qwen/Qwen3.5-4B-Base
+library_name: transformers
+model_name: Qwen3.5-4B-Base_overgeneralisation_splits_original_features_train_overgeneralisation_splits_original_features_test2
+tags:
+- generated_from_trainer
+- trl
+- sft
+licence: license
+---
+
+# Model Card for Qwen3.5-4B-Base_overgeneralisation_splits_original_features_train_overgeneralisation_splits_original_features_test2
+
+This model is a fine-tuned version of [Qwen/Qwen3.5-4B-Base](https://huggingface.co/Qwen/Qwen3.5-4B-Base).
+It has been trained using [TRL](https://github.com/huggingface/trl).
+
+## Quick start
+
+```python
+from transformers import pipeline
+
+question = "If you had a time machine, but could only go to the past or the future once and never return, which would you choose and why?"
+generator = pipeline("text-generation", model="None", device="cuda")
+output = generator([{"role": "user", "content": question}], max_new_tokens=128, return_full_text=False)[0]
+print(output["generated_text"])
+```
+
+## Training procedure
+
+[<img src="https://raw.githubusercontent.com/wandb/assets/main/wandb-github-badge-28.svg" alt="Visualize in Weights & Biases" width="150" height="24"/>](https://wandb.ai/katriin-kukk/Cross_lingual_morphological_generalization/runs/gtp4i6ta) 
+
+
+
+This model was trained with SFT.
+
+### Framework versions
+
+- TRL: 0.29.0
+- Transformers: 5.5.4
+- Pytorch: 2.10.0
+- Datasets: 4.6.1
+- Tokenizers: 0.22.2
+
+## Citations
+
+
+
+Cite TRL as:
+    
+```bibtex
+@software{vonwerra2020trl,
+  title   = {{TRL: Transformers Reinforcement Learning}},
+  author  = {von Werra, Leandro and Belkada, Younes and Tunstall, Lewis and Beeching, Edward and Thrush, Tristan and Lambert, Nathan and Huang, Shengyi and Rasul, Kashif and Gallouédec, Quentin},
+  license = {Apache-2.0},
+  url     = {https://github.com/huggingface/trl},
+  year    = {2020}
+}
+```

overgeneralisation_original_Estonian/Qwen3.5-4B-Base_overgeneralisation_splits_original_features_train_overgeneralisation_splits_original_features_test2/checkpoint-100/README.md

@@ -0,0 +1,209 @@
+---
+base_model: Qwen/Qwen3.5-4B-Base
+library_name: peft
+pipeline_tag: text-generation
+tags:
+- base_model:adapter:Qwen/Qwen3.5-4B-Base
+- lora
+- sft
+- transformers
+- trl
+---
+
+# Model Card for Model ID
+
+<!-- Provide a quick summary of what the model is/does. -->
+
+
+
+## Model Details
+
+### Model Description
+
+<!-- Provide a longer summary of what this model is. -->
+
+
+
+- **Developed by:** [More Information Needed]
+- **Funded by [optional]:** [More Information Needed]
+- **Shared by [optional]:** [More Information Needed]
+- **Model type:** [More Information Needed]
+- **Language(s) (NLP):** [More Information Needed]
+- **License:** [More Information Needed]
+- **Finetuned from model [optional]:** [More Information Needed]
+
+### Model Sources [optional]
+
+<!-- Provide the basic links for the model. -->
+
+- **Repository:** [More Information Needed]
+- **Paper [optional]:** [More Information Needed]
+- **Demo [optional]:** [More Information Needed]
+
+## Uses
+
+<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
+
+### Direct Use
+
+<!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
+
+[More Information Needed]
+
+### Downstream Use [optional]
+
+<!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->
+
+[More Information Needed]
+
+### Out-of-Scope Use
+
+<!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
+
+[More Information Needed]
+
+## Bias, Risks, and Limitations
+
+<!-- This section is meant to convey both technical and sociotechnical limitations. -->
+
+[More Information Needed]
+
+### Recommendations
+
+<!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->
+
+Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
+
+## How to Get Started with the Model
+
+Use the code below to get started with the model.
+
+[More Information Needed]
+
+## Training Details
+
+### Training Data
+
+<!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
+
+[More Information Needed]
+
+### Training Procedure
+
+<!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
+
+#### Preprocessing [optional]
+
+[More Information Needed]
+
+
+#### Training Hyperparameters
+
+- **Training regime:** [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->
+
+#### Speeds, Sizes, Times [optional]
+
+<!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
+
+[More Information Needed]
+
+## Evaluation
+
+<!-- This section describes the evaluation protocols and provides the results. -->
+
+### Testing Data, Factors & Metrics
+
+#### Testing Data
+
+<!-- This should link to a Dataset Card if possible. -->
+
+[More Information Needed]
+
+#### Factors
+
+<!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->
+
+[More Information Needed]
+
+#### Metrics
+
+<!-- These are the evaluation metrics being used, ideally with a description of why. -->
+
+[More Information Needed]
+
+### Results
+
+[More Information Needed]
+
+#### Summary
+
+
+
+## Model Examination [optional]
+
+<!-- Relevant interpretability work for the model goes here -->
+
+[More Information Needed]
+
+## Environmental Impact
+
+<!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly -->
+
+Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700).
+
+- **Hardware Type:** [More Information Needed]
+- **Hours used:** [More Information Needed]
+- **Cloud Provider:** [More Information Needed]
+- **Compute Region:** [More Information Needed]
+- **Carbon Emitted:** [More Information Needed]
+
+## Technical Specifications [optional]
+
+### Model Architecture and Objective
+
+[More Information Needed]
+
+### Compute Infrastructure
+
+[More Information Needed]
+
+#### Hardware
+
+[More Information Needed]
+
+#### Software
+
+[More Information Needed]
+
+## Citation [optional]
+
+<!-- If there is a paper or blog post introducing the model, the APA and Bibtex information for that should go in this section. -->
+
+**BibTeX:**
+
+[More Information Needed]
+
+**APA:**
+
+[More Information Needed]
+
+## Glossary [optional]
+
+<!-- If relevant, include terms and calculations in this section that can help readers understand the model or model card. -->
+
+[More Information Needed]
+
+## More Information [optional]
+
+[More Information Needed]
+
+## Model Card Authors [optional]
+
+[More Information Needed]
+
+## Model Card Contact
+
+[More Information Needed]
+### Framework versions
+
+- PEFT 0.18.1

overgeneralisation_original_Estonian/Qwen3.5-4B-Base_overgeneralisation_splits_original_features_train_overgeneralisation_splits_original_features_test2/checkpoint-100/adapter_config.json

@@ -0,0 +1,46 @@
+{
+  "alora_invocation_tokens": null,
+  "alpha_pattern": {},
+  "arrow_config": null,
+  "auto_mapping": null,
+  "base_model_name_or_path": "Qwen/Qwen3.5-4B-Base",
+  "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": 128,
+  "lora_bias": false,
+  "lora_dropout": 0.03828026524568501,
+  "megatron_config": null,
+  "megatron_core": "megatron.core",
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "peft_version": "0.18.1",
+  "qalora_group_size": 16,
+  "r": 64,
+  "rank_pattern": {},
+  "revision": null,
+  "target_modules": [
+    "gate_proj",
+    "v_proj",
+    "up_proj",
+    "down_proj",
+    "o_proj",
+    "q_proj",
+    "k_proj"
+  ],
+  "target_parameters": null,
+  "task_type": "CAUSAL_LM",
+  "trainable_token_indices": null,
+  "use_dora": false,
+  "use_qalora": false,
+  "use_rslora": false
+}

overgeneralisation_original_Estonian/Qwen3.5-4B-Base_overgeneralisation_splits_original_features_train_overgeneralisation_splits_original_features_test2/checkpoint-100/chat_template.jinja

@@ -0,0 +1,154 @@
+{%- set image_count = namespace(value=0) %}
+{%- set video_count = namespace(value=0) %}
+{%- macro render_content(content, do_vision_count, is_system_content=false) %}
+    {%- if content is string %}
+        {{- content }}
+    {%- elif content is iterable and content is not mapping %}
+        {%- for item in content %}
+            {%- if 'image' in item or 'image_url' in item or item.type == 'image' %}
+                {%- if is_system_content %}
+                    {{- raise_exception('System message cannot contain images.') }}
+                {%- endif %}
+                {%- if do_vision_count %}
+                    {%- set image_count.value = image_count.value + 1 %}
+                {%- endif %}
+                {%- if add_vision_id %}
+                    {{- 'Picture ' ~ image_count.value ~ ': ' }}
+                {%- endif %}
+                {{- '<|vision_start|><|image_pad|><|vision_end|>' }}
+            {%- elif 'video' in item or item.type == 'video' %}
+                {%- if is_system_content %}
+                    {{- raise_exception('System message cannot contain videos.') }}
+                {%- endif %}
+                {%- if do_vision_count %}
+                    {%- set video_count.value = video_count.value + 1 %}
+                {%- endif %}
+                {%- if add_vision_id %}
+                    {{- 'Video ' ~ video_count.value ~ ': ' }}
+                {%- endif %}
+                {{- '<|vision_start|><|video_pad|><|vision_end|>' }}
+            {%- elif 'text' in item %}
+                {{- item.text }}
+            {%- else %}
+                {{- raise_exception('Unexpected item type in content.') }}
+            {%- endif %}
+        {%- endfor %}
+    {%- elif content is none or content is undefined %}
+        {{- '' }}
+    {%- else %}
+        {{- raise_exception('Unexpected content type.') }}
+    {%- endif %}
+{%- endmacro %}
+{%- if not messages %}
+    {{- raise_exception('No messages provided.') }}
+{%- endif %}
+{%- if tools and tools is iterable and tools is not mapping %}
+    {{- '<|im_start|>system\n' }}
+    {{- "# Tools\n\nYou have access to the following functions:\n\n<tools>" }}
+    {%- for tool in tools %}
+        {{- "\n" }}
+        {{- tool | tojson }}
+    {%- endfor %}
+    {{- "\n</tools>" }}
+    {{- '\n\nIf you choose to call a function ONLY reply in the following format with NO suffix:\n\n<tool_call>\n<function=example_function_name>\n<parameter=example_parameter_1>\nvalue_1\n</parameter>\n<parameter=example_parameter_2>\nThis is the value for the second parameter\nthat can span\nmultiple lines\n</parameter>\n</function>\n</tool_call>\n\n<IMPORTANT>\nReminder:\n- Function calls MUST follow the specified format: an inner <function=...></function> block must be nested within <tool_call></tool_call> XML tags\n- Required parameters MUST be specified\n- You may provide optional reasoning for your function call in natural language BEFORE the function call, but NOT after\n- If there is no function call available, answer the question like normal with your current knowledge and do not tell the user about function calls\n</IMPORTANT>' }}
+    {%- if messages[0].role == 'system' %}
+        {%- set content = render_content(messages[0].content, false, true)|trim %}
+        {%- if content %}
+            {{- '\n\n' + content }}
+        {%- endif %}
+    {%- endif %}
+    {{- '<|im_end|>\n' }}
+{%- else %}
+    {%- if messages[0].role == 'system' %}
+        {%- set content = render_content(messages[0].content, false, true)|trim %}
+        {{- '<|im_start|>system\n' + content + '<|im_end|>\n' }}
+    {%- endif %}
+{%- endif %}
+{%- set ns = namespace(multi_step_tool=true, last_query_index=messages|length - 1) %}
+{%- for message in messages[::-1] %}
+    {%- set index = (messages|length - 1) - loop.index0 %}
+    {%- if ns.multi_step_tool and message.role == "user" %}
+        {%- set content = render_content(message.content, false)|trim %}
+        {%- if not(content.startswith('<tool_response>') and content.endswith('</tool_response>')) %}
+            {%- set ns.multi_step_tool = false %}
+            {%- set ns.last_query_index = index %}
+        {%- endif %}
+    {%- endif %}
+{%- endfor %}
+{%- if ns.multi_step_tool %}
+    {{- raise_exception('No user query found in messages.') }}
+{%- endif %}
+{%- for message in messages %}
+    {%- set content = render_content(message.content, true)|trim %}
+    {%- if message.role == "system" %}
+        {%- if not loop.first %}
+            {{- raise_exception('System message must be at the beginning.') }}
+        {%- endif %}
+    {%- elif message.role == "user" %}
+        {{- '<|im_start|>' + message.role + '\n' + content + '<|im_end|>' + '\n' }}
+    {%- elif message.role == "assistant" %}
+        {%- set reasoning_content = '' %}
+        {%- if message.reasoning_content is string %}
+            {%- set reasoning_content = message.reasoning_content %}
+        {%- else %}
+            {%- if '</think>' in content %}
+                {%- set reasoning_content = content.split('</think>')[0].rstrip('\n').split('<think>')[-1].lstrip('\n') %}
+                {%- set content = content.split('</think>')[-1].lstrip('\n') %}
+            {%- endif %}
+        {%- endif %}
+        {%- set reasoning_content = reasoning_content|trim %}
+        {%- if loop.index0 > ns.last_query_index %}
+            {{- '<|im_start|>' + message.role + '\n<think>\n' + reasoning_content + '\n</think>\n\n' + content }}
+        {%- else %}
+            {{- '<|im_start|>' + message.role + '\n' + content }}
+        {%- endif %}
+        {%- if message.tool_calls and message.tool_calls is iterable and message.tool_calls is not mapping %}
+            {%- for tool_call in message.tool_calls %}
+                {%- if tool_call.function is defined %}
+                    {%- set tool_call = tool_call.function %}
+                {%- endif %}
+                {%- if loop.first %}
+                    {%- if content|trim %}
+                        {{- '\n\n<tool_call>\n<function=' + tool_call.name + '>\n' }}
+                    {%- else %}
+                        {{- '<tool_call>\n<function=' + tool_call.name + '>\n' }}
+                    {%- endif %}
+                {%- else %}
+                    {{- '\n<tool_call>\n<function=' + tool_call.name + '>\n' }}
+                {%- endif %}
+                {%- if tool_call.arguments is defined %}
+                    {%- for args_name, args_value in tool_call.arguments|items %}
+                        {{- '<parameter=' + args_name + '>\n' }}
+                        {%- set args_value = args_value | tojson | safe if args_value is mapping or (args_value is sequence and args_value is not string) else args_value | string %}
+                        {{- args_value }}
+                        {{- '\n</parameter>\n' }}
+                    {%- endfor %}
+                {%- endif %}
+                {{- '</function>\n</tool_call>' }}
+            {%- endfor %}
+        {%- endif %}
+        {{- '<|im_end|>\n' }}
+    {%- elif message.role == "tool" %}
+        {%- if loop.previtem and loop.previtem.role != "tool" %}
+            {{- '<|im_start|>user' }}
+        {%- endif %}
+        {{- '\n<tool_response>\n' }}
+        {{- content }}
+        {{- '\n</tool_response>' }}
+        {%- if not loop.last and loop.nextitem.role != "tool" %}
+            {{- '<|im_end|>\n' }}
+        {%- elif loop.last %}
+            {{- '<|im_end|>\n' }}
+        {%- endif %}
+    {%- else %}
+        {{- raise_exception('Unexpected message role.') }}
+    {%- endif %}
+{%- endfor %}
+{%- if add_generation_prompt %}
+    {{- '<|im_start|>assistant\n' }}
+    {%- if enable_thinking is defined and enable_thinking is false %}
+        {{- '<think>\n\n</think>\n\n' }}
+    {%- else %}
+        {{- '<think>\n' }}
+    {%- endif %}
+{%- endif %}

overgeneralisation_original_Estonian/Qwen3.5-4B-Base_overgeneralisation_splits_original_features_train_overgeneralisation_splits_original_features_test2/checkpoint-100/tokenizer_config.json

@@ -0,0 +1,31 @@
+{
+  "add_prefix_space": false,
+  "audio_bos_token": "<|audio_start|>",
+  "audio_eos_token": "<|audio_end|>",
+  "audio_token": "<|audio_pad|>",
+  "backend": "tokenizers",
+  "bos_token": null,
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<|endoftext|>",
+  "errors": "replace",
+  "image_token": "<|image_pad|>",
+  "is_local": false,
+  "model_max_length": 262144,
+  "model_specific_special_tokens": {
+    "audio_bos_token": "<|audio_start|>",
+    "audio_eos_token": "<|audio_end|>",
+    "audio_token": "<|audio_pad|>",
+    "image_token": "<|image_pad|>",
+    "video_token": "<|video_pad|>",
+    "vision_bos_token": "<|vision_start|>",
+    "vision_eos_token": "<|vision_end|>"
+  },
+  "pad_token": "<|endoftext|>",
+  "pretokenize_regex": "(?i:'s|'t|'re|'ve|'m|'ll|'d)|[^\\r\\n\\p{L}\\p{N}]?[\\p{L}\\p{M}]+|\\p{N}| ?[^\\s\\p{L}\\p{M}\\p{N}]+[\\r\\n]*|\\s*[\\r\\n]+|\\s+(?!\\S)|\\s+",
+  "split_special_tokens": false,
+  "tokenizer_class": "TokenizersBackend",
+  "unk_token": null,
+  "video_token": "<|video_pad|>",
+  "vision_bos_token": "<|vision_start|>",
+  "vision_eos_token": "<|vision_end|>"
+}

overgeneralisation_original_Estonian/Qwen3.5-4B-Base_overgeneralisation_splits_original_features_train_overgeneralisation_splits_original_features_test2/checkpoint-100/trainer_state.json

@@ -0,0 +1,139 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.24140012070006034,
+  "eval_steps": 20,
+  "global_step": 100,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "entropy": 2.172999547421932,
+      "epoch": 0.04828002414001207,
+      "grad_norm": 2.5071208477020264,
+      "learning_rate": 1.7227585487708594e-05,
+      "loss": 1.9347333908081055,
+      "mean_token_accuracy": 0.6086816020309925,
+      "num_tokens": 47778.0,
+      "step": 20
+    },
+    {
+      "epoch": 0.04828002414001207,
+      "eval_entropy": 1.6909754791956269,
+      "eval_loss": 1.4728385210037231,
+      "eval_mean_token_accuracy": 0.6773593797442619,
+      "eval_num_tokens": 47778.0,
+      "eval_runtime": 90.1265,
+      "eval_samples_per_second": 15.756,
+      "eval_steps_per_second": 1.975,
+      "step": 20
+    },
+    {
+      "entropy": 1.2014626950025558,
+      "epoch": 0.09656004828002414,
+      "grad_norm": 1.5074543952941895,
+      "learning_rate": 3.5361886001086065e-05,
+      "loss": 1.0867146492004394,
+      "mean_token_accuracy": 0.7285927519202232,
+      "num_tokens": 95981.0,
+      "step": 40
+    },
+    {
+      "epoch": 0.09656004828002414,
+      "eval_entropy": 1.0304168352250302,
+      "eval_loss": 0.9211422204971313,
+      "eval_mean_token_accuracy": 0.7613122848312507,
+      "eval_num_tokens": 95981.0,
+      "eval_runtime": 88.6338,
+      "eval_samples_per_second": 16.021,
+      "eval_steps_per_second": 2.008,
+      "step": 40
+    },
+    {
+      "entropy": 0.9454198732972146,
+      "epoch": 0.14484007242003621,
+      "grad_norm": 1.7578108310699463,
+      "learning_rate": 5.349618651446354e-05,
+      "loss": 0.8447297096252442,
+      "mean_token_accuracy": 0.7723424732685089,
+      "num_tokens": 142784.0,
+      "step": 60
+    },
+    {
+      "epoch": 0.14484007242003621,
+      "eval_entropy": 0.8647834474451086,
+      "eval_loss": 0.8240737318992615,
+      "eval_mean_token_accuracy": 0.7821227014064789,
+      "eval_num_tokens": 142784.0,
+      "eval_runtime": 88.0135,
+      "eval_samples_per_second": 16.134,
+      "eval_steps_per_second": 2.022,
+      "step": 60
+    },
+    {
+      "entropy": 0.8963550612330436,
+      "epoch": 0.19312009656004828,
+      "grad_norm": 1.1464931964874268,
+      "learning_rate": 7.1630487027841e-05,
+      "loss": 0.7979836463928223,
+      "mean_token_accuracy": 0.7822489373385906,
+      "num_tokens": 187338.0,
+      "step": 80
+    },
+    {
+      "epoch": 0.19312009656004828,
+      "eval_entropy": 0.8663296133614657,
+      "eval_loss": 0.7752988338470459,
+      "eval_mean_token_accuracy": 0.7885957012015782,
+      "eval_num_tokens": 187338.0,
+      "eval_runtime": 88.284,
+      "eval_samples_per_second": 16.084,
+      "eval_steps_per_second": 2.016,
+      "step": 80
+    },
+    {
+      "entropy": 0.8265103206038475,
+      "epoch": 0.24140012070006034,
+      "grad_norm": 1.0106662511825562,
+      "learning_rate": 8.976478754121848e-05,
+      "loss": 0.7297060012817382,
+      "mean_token_accuracy": 0.7955138415098191,
+      "num_tokens": 236038.0,
+      "step": 100
+    },
+    {
+      "epoch": 0.24140012070006034,
+      "eval_entropy": 0.8122169536151244,
+      "eval_loss": 0.754906177520752,
+      "eval_mean_token_accuracy": 0.7917155255092664,
+      "eval_num_tokens": 236038.0,
+      "eval_runtime": 87.5754,
+      "eval_samples_per_second": 16.215,
+      "eval_steps_per_second": 2.033,
+      "step": 100
+    }
+  ],
+  "logging_steps": 20,
+  "max_steps": 4150,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 10,
+  "save_steps": 20,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 9897324005793792.0,
+  "train_batch_size": 4,
+  "trial_name": null,
+  "trial_params": null
+}

overgeneralisation_original_Estonian/Qwen3.5-4B-Base_overgeneralisation_splits_original_features_train_overgeneralisation_splits_original_features_test2/checkpoint-1000/README.md

@@ -0,0 +1,209 @@
+---
+base_model: Qwen/Qwen3.5-4B-Base
+library_name: peft
+pipeline_tag: text-generation
+tags:
+- base_model:adapter:Qwen/Qwen3.5-4B-Base
+- lora
+- sft
+- transformers
+- trl
+---
+
+# Model Card for Model ID
+
+<!-- Provide a quick summary of what the model is/does. -->
+
+
+
+## Model Details
+
+### Model Description
+
+<!-- Provide a longer summary of what this model is. -->
+
+
+
+- **Developed by:** [More Information Needed]
+- **Funded by [optional]:** [More Information Needed]
+- **Shared by [optional]:** [More Information Needed]
+- **Model type:** [More Information Needed]
+- **Language(s) (NLP):** [More Information Needed]
+- **License:** [More Information Needed]
+- **Finetuned from model [optional]:** [More Information Needed]
+
+### Model Sources [optional]
+
+<!-- Provide the basic links for the model. -->
+
+- **Repository:** [More Information Needed]
+- **Paper [optional]:** [More Information Needed]
+- **Demo [optional]:** [More Information Needed]
+
+## Uses
+
+<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
+
+### Direct Use
+
+<!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
+
+[More Information Needed]
+
+### Downstream Use [optional]
+
+<!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->
+
+[More Information Needed]
+
+### Out-of-Scope Use
+
+<!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
+
+[More Information Needed]
+
+## Bias, Risks, and Limitations
+
+<!-- This section is meant to convey both technical and sociotechnical limitations. -->
+
+[More Information Needed]
+
+### Recommendations
+
+<!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->
+
+Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
+
+## How to Get Started with the Model
+
+Use the code below to get started with the model.
+
+[More Information Needed]
+
+## Training Details
+
+### Training Data
+
+<!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
+
+[More Information Needed]
+
+### Training Procedure
+
+<!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
+
+#### Preprocessing [optional]
+
+[More Information Needed]
+
+
+#### Training Hyperparameters
+
+- **Training regime:** [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->
+
+#### Speeds, Sizes, Times [optional]
+
+<!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
+
+[More Information Needed]
+
+## Evaluation
+
+<!-- This section describes the evaluation protocols and provides the results. -->
+
+### Testing Data, Factors & Metrics
+
+#### Testing Data
+
+<!-- This should link to a Dataset Card if possible. -->
+
+[More Information Needed]
+
+#### Factors
+
+<!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->
+
+[More Information Needed]
+
+#### Metrics
+
+<!-- These are the evaluation metrics being used, ideally with a description of why. -->
+
+[More Information Needed]
+
+### Results
+
+[More Information Needed]
+
+#### Summary
+
+
+
+## Model Examination [optional]
+
+<!-- Relevant interpretability work for the model goes here -->
+
+[More Information Needed]
+
+## Environmental Impact
+
+<!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly -->
+
+Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700).
+
+- **Hardware Type:** [More Information Needed]
+- **Hours used:** [More Information Needed]
+- **Cloud Provider:** [More Information Needed]
+- **Compute Region:** [More Information Needed]
+- **Carbon Emitted:** [More Information Needed]
+
+## Technical Specifications [optional]
+
+### Model Architecture and Objective
+
+[More Information Needed]
+
+### Compute Infrastructure
+
+[More Information Needed]
+
+#### Hardware
+
+[More Information Needed]
+
+#### Software
+
+[More Information Needed]
+
+## Citation [optional]
+
+<!-- If there is a paper or blog post introducing the model, the APA and Bibtex information for that should go in this section. -->
+
+**BibTeX:**
+
+[More Information Needed]
+
+**APA:**
+
+[More Information Needed]
+
+## Glossary [optional]
+
+<!-- If relevant, include terms and calculations in this section that can help readers understand the model or model card. -->
+
+[More Information Needed]
+
+## More Information [optional]
+
+[More Information Needed]
+
+## Model Card Authors [optional]
+
+[More Information Needed]
+
+## Model Card Contact
+
+[More Information Needed]
+### Framework versions
+
+- PEFT 0.18.1

overgeneralisation_original_Estonian/Qwen3.5-4B-Base_overgeneralisation_splits_original_features_train_overgeneralisation_splits_original_features_test2/checkpoint-1000/adapter_config.json

@@ -0,0 +1,46 @@
+{
+  "alora_invocation_tokens": null,
+  "alpha_pattern": {},
+  "arrow_config": null,
+  "auto_mapping": null,
+  "base_model_name_or_path": "Qwen/Qwen3.5-4B-Base",
+  "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": 128,
+  "lora_bias": false,
+  "lora_dropout": 0.03828026524568501,
+  "megatron_config": null,
+  "megatron_core": "megatron.core",
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "peft_version": "0.18.1",
+  "qalora_group_size": 16,
+  "r": 64,
+  "rank_pattern": {},
+  "revision": null,
+  "target_modules": [
+    "gate_proj",
+    "v_proj",
+    "up_proj",
+    "down_proj",
+    "o_proj",
+    "q_proj",
+    "k_proj"
+  ],
+  "target_parameters": null,
+  "task_type": "CAUSAL_LM",
+  "trainable_token_indices": null,
+  "use_dora": false,
+  "use_qalora": false,
+  "use_rslora": false
+}

overgeneralisation_original_Estonian/Qwen3.5-4B-Base_overgeneralisation_splits_original_features_train_overgeneralisation_splits_original_features_test2/checkpoint-1000/chat_template.jinja

@@ -0,0 +1,154 @@
+{%- set image_count = namespace(value=0) %}
+{%- set video_count = namespace(value=0) %}
+{%- macro render_content(content, do_vision_count, is_system_content=false) %}
+    {%- if content is string %}
+        {{- content }}
+    {%- elif content is iterable and content is not mapping %}
+        {%- for item in content %}
+            {%- if 'image' in item or 'image_url' in item or item.type == 'image' %}
+                {%- if is_system_content %}
+                    {{- raise_exception('System message cannot contain images.') }}
+                {%- endif %}
+                {%- if do_vision_count %}
+                    {%- set image_count.value = image_count.value + 1 %}
+                {%- endif %}
+                {%- if add_vision_id %}
+                    {{- 'Picture ' ~ image_count.value ~ ': ' }}
+                {%- endif %}
+                {{- '<|vision_start|><|image_pad|><|vision_end|>' }}
+            {%- elif 'video' in item or item.type == 'video' %}
+                {%- if is_system_content %}
+                    {{- raise_exception('System message cannot contain videos.') }}
+                {%- endif %}
+                {%- if do_vision_count %}
+                    {%- set video_count.value = video_count.value + 1 %}
+                {%- endif %}
+                {%- if add_vision_id %}
+                    {{- 'Video ' ~ video_count.value ~ ': ' }}
+                {%- endif %}
+                {{- '<|vision_start|><|video_pad|><|vision_end|>' }}
+            {%- elif 'text' in item %}
+                {{- item.text }}
+            {%- else %}
+                {{- raise_exception('Unexpected item type in content.') }}
+            {%- endif %}
+        {%- endfor %}
+    {%- elif content is none or content is undefined %}
+        {{- '' }}
+    {%- else %}
+        {{- raise_exception('Unexpected content type.') }}
+    {%- endif %}
+{%- endmacro %}
+{%- if not messages %}
+    {{- raise_exception('No messages provided.') }}
+{%- endif %}
+{%- if tools and tools is iterable and tools is not mapping %}
+    {{- '<|im_start|>system\n' }}
+    {{- "# Tools\n\nYou have access to the following functions:\n\n<tools>" }}
+    {%- for tool in tools %}
+        {{- "\n" }}
+        {{- tool | tojson }}
+    {%- endfor %}
+    {{- "\n</tools>" }}
+    {{- '\n\nIf you choose to call a function ONLY reply in the following format with NO suffix:\n\n<tool_call>\n<function=example_function_name>\n<parameter=example_parameter_1>\nvalue_1\n</parameter>\n<parameter=example_parameter_2>\nThis is the value for the second parameter\nthat can span\nmultiple lines\n</parameter>\n</function>\n</tool_call>\n\n<IMPORTANT>\nReminder:\n- Function calls MUST follow the specified format: an inner <function=...></function> block must be nested within <tool_call></tool_call> XML tags\n- Required parameters MUST be specified\n- You may provide optional reasoning for your function call in natural language BEFORE the function call, but NOT after\n- If there is no function call available, answer the question like normal with your current knowledge and do not tell the user about function calls\n</IMPORTANT>' }}
+    {%- if messages[0].role == 'system' %}
+        {%- set content = render_content(messages[0].content, false, true)|trim %}
+        {%- if content %}
+            {{- '\n\n' + content }}
+        {%- endif %}
+    {%- endif %}
+    {{- '<|im_end|>\n' }}
+{%- else %}
+    {%- if messages[0].role == 'system' %}
+        {%- set content = render_content(messages[0].content, false, true)|trim %}
+        {{- '<|im_start|>system\n' + content + '<|im_end|>\n' }}
+    {%- endif %}
+{%- endif %}
+{%- set ns = namespace(multi_step_tool=true, last_query_index=messages|length - 1) %}
+{%- for message in messages[::-1] %}
+    {%- set index = (messages|length - 1) - loop.index0 %}
+    {%- if ns.multi_step_tool and message.role == "user" %}
+        {%- set content = render_content(message.content, false)|trim %}
+        {%- if not(content.startswith('<tool_response>') and content.endswith('</tool_response>')) %}
+            {%- set ns.multi_step_tool = false %}
+            {%- set ns.last_query_index = index %}
+        {%- endif %}
+    {%- endif %}
+{%- endfor %}
+{%- if ns.multi_step_tool %}
+    {{- raise_exception('No user query found in messages.') }}
+{%- endif %}
+{%- for message in messages %}
+    {%- set content = render_content(message.content, true)|trim %}
+    {%- if message.role == "system" %}
+        {%- if not loop.first %}
+            {{- raise_exception('System message must be at the beginning.') }}
+        {%- endif %}
+    {%- elif message.role == "user" %}
+        {{- '<|im_start|>' + message.role + '\n' + content + '<|im_end|>' + '\n' }}
+    {%- elif message.role == "assistant" %}
+        {%- set reasoning_content = '' %}
+        {%- if message.reasoning_content is string %}
+            {%- set reasoning_content = message.reasoning_content %}
+        {%- else %}
+            {%- if '</think>' in content %}
+                {%- set reasoning_content = content.split('</think>')[0].rstrip('\n').split('<think>')[-1].lstrip('\n') %}
+                {%- set content = content.split('</think>')[-1].lstrip('\n') %}
+            {%- endif %}
+        {%- endif %}
+        {%- set reasoning_content = reasoning_content|trim %}
+        {%- if loop.index0 > ns.last_query_index %}
+            {{- '<|im_start|>' + message.role + '\n<think>\n' + reasoning_content + '\n</think>\n\n' + content }}
+        {%- else %}
+            {{- '<|im_start|>' + message.role + '\n' + content }}
+        {%- endif %}
+        {%- if message.tool_calls and message.tool_calls is iterable and message.tool_calls is not mapping %}
+            {%- for tool_call in message.tool_calls %}
+                {%- if tool_call.function is defined %}
+                    {%- set tool_call = tool_call.function %}
+                {%- endif %}
+                {%- if loop.first %}
+                    {%- if content|trim %}
+                        {{- '\n\n<tool_call>\n<function=' + tool_call.name + '>\n' }}
+                    {%- else %}
+                        {{- '<tool_call>\n<function=' + tool_call.name + '>\n' }}
+                    {%- endif %}
+                {%- else %}
+                    {{- '\n<tool_call>\n<function=' + tool_call.name + '>\n' }}
+                {%- endif %}
+                {%- if tool_call.arguments is defined %}
+                    {%- for args_name, args_value in tool_call.arguments|items %}
+                        {{- '<parameter=' + args_name + '>\n' }}
+                        {%- set args_value = args_value | tojson | safe if args_value is mapping or (args_value is sequence and args_value is not string) else args_value | string %}
+                        {{- args_value }}
+                        {{- '\n</parameter>\n' }}
+                    {%- endfor %}
+                {%- endif %}
+                {{- '</function>\n</tool_call>' }}
+            {%- endfor %}
+        {%- endif %}
+        {{- '<|im_end|>\n' }}
+    {%- elif message.role == "tool" %}
+        {%- if loop.previtem and loop.previtem.role != "tool" %}
+            {{- '<|im_start|>user' }}
+        {%- endif %}
+        {{- '\n<tool_response>\n' }}
+        {{- content }}
+        {{- '\n</tool_response>' }}
+        {%- if not loop.last and loop.nextitem.role != "tool" %}
+            {{- '<|im_end|>\n' }}
+        {%- elif loop.last %}
+            {{- '<|im_end|>\n' }}
+        {%- endif %}
+    {%- else %}
+        {{- raise_exception('Unexpected message role.') }}
+    {%- endif %}
+{%- endfor %}
+{%- if add_generation_prompt %}
+    {{- '<|im_start|>assistant\n' }}
+    {%- if enable_thinking is defined and enable_thinking is false %}
+        {{- '<think>\n\n</think>\n\n' }}
+    {%- else %}
+        {{- '<think>\n' }}
+    {%- endif %}
+{%- endif %}

overgeneralisation_original_Estonian/Qwen3.5-4B-Base_overgeneralisation_splits_original_features_train_overgeneralisation_splits_original_features_test2/checkpoint-1000/tokenizer_config.json

@@ -0,0 +1,31 @@
+{
+  "add_prefix_space": false,
+  "audio_bos_token": "<|audio_start|>",
+  "audio_eos_token": "<|audio_end|>",
+  "audio_token": "<|audio_pad|>",
+  "backend": "tokenizers",
+  "bos_token": null,
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<|endoftext|>",
+  "errors": "replace",
+  "image_token": "<|image_pad|>",
+  "is_local": false,
+  "model_max_length": 262144,
+  "model_specific_special_tokens": {
+    "audio_bos_token": "<|audio_start|>",
+    "audio_eos_token": "<|audio_end|>",
+    "audio_token": "<|audio_pad|>",
+    "image_token": "<|image_pad|>",
+    "video_token": "<|video_pad|>",
+    "vision_bos_token": "<|vision_start|>",
+    "vision_eos_token": "<|vision_end|>"
+  },
+  "pad_token": "<|endoftext|>",
+  "pretokenize_regex": "(?i:'s|'t|'re|'ve|'m|'ll|'d)|[^\\r\\n\\p{L}\\p{N}]?[\\p{L}\\p{M}]+|\\p{N}| ?[^\\s\\p{L}\\p{M}\\p{N}]+[\\r\\n]*|\\s*[\\r\\n]+|\\s+(?!\\S)|\\s+",
+  "split_special_tokens": false,
+  "tokenizer_class": "TokenizersBackend",
+  "unk_token": null,
+  "video_token": "<|video_pad|>",
+  "vision_bos_token": "<|vision_start|>",
+  "vision_eos_token": "<|vision_end|>"
+}

overgeneralisation_original_Estonian/Qwen3.5-4B-Base_overgeneralisation_splits_original_features_train_overgeneralisation_splits_original_features_test2/checkpoint-1000/trainer_state.json

@@ -0,0 +1,1084 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 2.4103802051901027,
+  "eval_steps": 20,
+  "global_step": 1000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "entropy": 2.172999547421932,
+      "epoch": 0.04828002414001207,
+      "grad_norm": 2.5071208477020264,
+      "learning_rate": 1.7227585487708594e-05,
+      "loss": 1.9347333908081055,
+      "mean_token_accuracy": 0.6086816020309925,
+      "num_tokens": 47778.0,
+      "step": 20
+    },
+    {
+      "epoch": 0.04828002414001207,
+      "eval_entropy": 1.6909754791956269,
+      "eval_loss": 1.4728385210037231,
+      "eval_mean_token_accuracy": 0.6773593797442619,
+      "eval_num_tokens": 47778.0,
+      "eval_runtime": 90.1265,
+      "eval_samples_per_second": 15.756,
+      "eval_steps_per_second": 1.975,
+      "step": 20
+    },
+    {
+      "entropy": 1.2014626950025558,
+      "epoch": 0.09656004828002414,
+      "grad_norm": 1.5074543952941895,
+      "learning_rate": 3.5361886001086065e-05,
+      "loss": 1.0867146492004394,
+      "mean_token_accuracy": 0.7285927519202232,
+      "num_tokens": 95981.0,
+      "step": 40
+    },
+    {
+      "epoch": 0.09656004828002414,
+      "eval_entropy": 1.0304168352250302,
+      "eval_loss": 0.9211422204971313,
+      "eval_mean_token_accuracy": 0.7613122848312507,
+      "eval_num_tokens": 95981.0,
+      "eval_runtime": 88.6338,
+      "eval_samples_per_second": 16.021,
+      "eval_steps_per_second": 2.008,
+      "step": 40
+    },
+    {
+      "entropy": 0.9454198732972146,
+      "epoch": 0.14484007242003621,
+      "grad_norm": 1.7578108310699463,
+      "learning_rate": 5.349618651446354e-05,
+      "loss": 0.8447297096252442,
+      "mean_token_accuracy": 0.7723424732685089,
+      "num_tokens": 142784.0,
+      "step": 60
+    },
+    {
+      "epoch": 0.14484007242003621,
+      "eval_entropy": 0.8647834474451086,
+      "eval_loss": 0.8240737318992615,
+      "eval_mean_token_accuracy": 0.7821227014064789,
+      "eval_num_tokens": 142784.0,
+      "eval_runtime": 88.0135,
+      "eval_samples_per_second": 16.134,
+      "eval_steps_per_second": 2.022,
+      "step": 60
+    },
+    {
+      "entropy": 0.8963550612330436,
+      "epoch": 0.19312009656004828,
+      "grad_norm": 1.1464931964874268,
+      "learning_rate": 7.1630487027841e-05,
+      "loss": 0.7979836463928223,
+      "mean_token_accuracy": 0.7822489373385906,
+      "num_tokens": 187338.0,
+      "step": 80
+    },
+    {
+      "epoch": 0.19312009656004828,
+      "eval_entropy": 0.8663296133614657,
+      "eval_loss": 0.7752988338470459,
+      "eval_mean_token_accuracy": 0.7885957012015782,
+      "eval_num_tokens": 187338.0,
+      "eval_runtime": 88.284,
+      "eval_samples_per_second": 16.084,
+      "eval_steps_per_second": 2.016,
+      "step": 80
+    },
+    {
+      "entropy": 0.8265103206038475,
+      "epoch": 0.24140012070006034,
+      "grad_norm": 1.0106662511825562,
+      "learning_rate": 8.976478754121848e-05,
+      "loss": 0.7297060012817382,
+      "mean_token_accuracy": 0.7955138415098191,
+      "num_tokens": 236038.0,
+      "step": 100
+    },
+    {
+      "epoch": 0.24140012070006034,
+      "eval_entropy": 0.8122169536151244,
+      "eval_loss": 0.754906177520752,
+      "eval_mean_token_accuracy": 0.7917155255092664,
+      "eval_num_tokens": 236038.0,
+      "eval_runtime": 87.5754,
+      "eval_samples_per_second": 16.215,
+      "eval_steps_per_second": 2.033,
+      "step": 100
+    },
+    {
+      "entropy": 0.8098392844200134,
+      "epoch": 0.28968014484007243,
+      "grad_norm": 1.2238041162490845,
+      "learning_rate": 0.00010789908805459595,
+      "loss": 0.7201489448547364,
+      "mean_token_accuracy": 0.7975016921758652,
+      "num_tokens": 282918.0,
+      "step": 120
+    },
+    {
+      "epoch": 0.28968014484007243,
+      "eval_entropy": 0.8418726462326692,
+      "eval_loss": 0.7393178939819336,
+      "eval_mean_token_accuracy": 0.7936771585700217,
+      "eval_num_tokens": 282918.0,
+      "eval_runtime": 87.1597,
+      "eval_samples_per_second": 16.292,
+      "eval_steps_per_second": 2.042,
+      "step": 120
+    },
+    {
+      "entropy": 0.8080591082572937,
+      "epoch": 0.33796016898008446,
+      "grad_norm": 1.011591911315918,
+      "learning_rate": 0.00012603338856797343,
+      "loss": 0.7147697925567627,
+      "mean_token_accuracy": 0.7947175532579422,
+      "num_tokens": 331305.0,
+      "step": 140
+    },
+    {
+      "epoch": 0.33796016898008446,
+      "eval_entropy": 0.7894941431083037,
+      "eval_loss": 0.720219075679779,
+      "eval_mean_token_accuracy": 0.801899236239744,
+      "eval_num_tokens": 331305.0,
+      "eval_runtime": 87.1493,
+      "eval_samples_per_second": 16.294,
+      "eval_steps_per_second": 2.042,
+      "step": 140
+    },
+    {
+      "entropy": 0.7957205109298229,
+      "epoch": 0.38624019312009655,
+      "grad_norm": 0.8481830954551697,
+      "learning_rate": 0.00014416768908135088,
+      "loss": 0.7050027847290039,
+      "mean_token_accuracy": 0.7995368830859662,
+      "num_tokens": 375876.0,
+      "step": 160
+    },
+    {
+      "epoch": 0.38624019312009655,
+      "eval_entropy": 0.7746140591883928,
+      "eval_loss": 0.7175475358963013,
+      "eval_mean_token_accuracy": 0.7994848955213354,
+      "eval_num_tokens": 375876.0,
+      "eval_runtime": 86.655,
+      "eval_samples_per_second": 16.387,
+      "eval_steps_per_second": 2.054,
+      "step": 160
+    },
+    {
+      "entropy": 0.779565304517746,
+      "epoch": 0.43452021726010864,
+      "grad_norm": 1.032624363899231,
+      "learning_rate": 0.00016230198959472835,
+      "loss": 0.6978332042694092,
+      "mean_token_accuracy": 0.8015618294477462,
+      "num_tokens": 423913.0,
+      "step": 180
+    },
+    {
+      "epoch": 0.43452021726010864,
+      "eval_entropy": 0.7455583871080634,
+      "eval_loss": 0.7118850350379944,
+      "eval_mean_token_accuracy": 0.7999682195400923,
+      "eval_num_tokens": 423913.0,
+      "eval_runtime": 86.7842,
+      "eval_samples_per_second": 16.362,
+      "eval_steps_per_second": 2.051,
+      "step": 180
+    },
+    {
+      "entropy": 0.7810172818601131,
+      "epoch": 0.4828002414001207,
+      "grad_norm": 0.9276663064956665,
+      "learning_rate": 0.00018043629010810582,
+      "loss": 0.6937861442565918,
+      "mean_token_accuracy": 0.802893053740263,
+      "num_tokens": 472395.0,
+      "step": 200
+    },
+    {
+      "epoch": 0.4828002414001207,
+      "eval_entropy": 0.7650193290764027,
+      "eval_loss": 0.7020567059516907,
+      "eval_mean_token_accuracy": 0.8045302153973097,
+      "eval_num_tokens": 472395.0,
+      "eval_runtime": 88.326,
+      "eval_samples_per_second": 16.077,
+      "eval_steps_per_second": 2.015,
+      "step": 200
+    },
+    {
+      "entropy": 0.7862060204148292,
+      "epoch": 0.5310802655401328,
+      "grad_norm": 0.8195295929908752,
+      "learning_rate": 0.0001985705906214833,
+      "loss": 0.6878782272338867,
+      "mean_token_accuracy": 0.8027168907225132,
+      "num_tokens": 517049.0,
+      "step": 220
+    },
+    {
+      "epoch": 0.5310802655401328,
+      "eval_entropy": 0.7656565583154057,
+      "eval_loss": 0.6973932981491089,
+      "eval_mean_token_accuracy": 0.8047967278555538,
+      "eval_num_tokens": 517049.0,
+      "eval_runtime": 87.7698,
+      "eval_samples_per_second": 16.179,
+      "eval_steps_per_second": 2.028,
+      "step": 220
+    },
+    {
+      "entropy": 0.7611534461379051,
+      "epoch": 0.5793602896801449,
+      "grad_norm": 0.7499712109565735,
+      "learning_rate": 0.00021670489113486077,
+      "loss": 0.681937837600708,
+      "mean_token_accuracy": 0.8041978515684605,
+      "num_tokens": 562836.0,
+      "step": 240
+    },
+    {
+      "epoch": 0.5793602896801449,
+      "eval_entropy": 0.7599442380197933,
+      "eval_loss": 0.7002046704292297,
+      "eval_mean_token_accuracy": 0.8051861517884759,
+      "eval_num_tokens": 562836.0,
+      "eval_runtime": 87.9239,
+      "eval_samples_per_second": 16.15,
+      "eval_steps_per_second": 2.024,
+      "step": 240
+    },
+    {
+      "entropy": 0.7800783462822437,
+      "epoch": 0.627640313820157,
+      "grad_norm": 0.8645269274711609,
+      "learning_rate": 0.00023483919164823825,
+      "loss": 0.6954500675201416,
+      "mean_token_accuracy": 0.8031484372913837,
+      "num_tokens": 606612.0,
+      "step": 260
+    },
+    {
+      "epoch": 0.627640313820157,
+      "eval_entropy": 0.7498679090751691,
+      "eval_loss": 0.7046149373054504,
+      "eval_mean_token_accuracy": 0.802388215667746,
+      "eval_num_tokens": 606612.0,
+      "eval_runtime": 88.2468,
+      "eval_samples_per_second": 16.091,
+      "eval_steps_per_second": 2.017,
+      "step": 260
+    },
+    {
+      "entropy": 0.7586533665657044,
+      "epoch": 0.6759203379601689,
+      "grad_norm": 0.741596519947052,
+      "learning_rate": 0.0002529734921616157,
+      "loss": 0.690476369857788,
+      "mean_token_accuracy": 0.8046923100948333,
+      "num_tokens": 655920.0,
+      "step": 280
+    },
+    {
+      "epoch": 0.6759203379601689,
+      "eval_entropy": 0.7714274066887544,
+      "eval_loss": 0.6992160677909851,
+      "eval_mean_token_accuracy": 0.8042646482419432,
+      "eval_num_tokens": 655920.0,
+      "eval_runtime": 88.1754,
+      "eval_samples_per_second": 16.104,
+      "eval_steps_per_second": 2.019,
+      "step": 280
+    },
+    {
+      "entropy": 0.7585848286747933,
+      "epoch": 0.724200362100181,
+      "grad_norm": 1.0778124332427979,
+      "learning_rate": 0.00027110779267499317,
+      "loss": 0.6790409564971924,
+      "mean_token_accuracy": 0.8060351841151714,
+      "num_tokens": 702638.0,
+      "step": 300
+    },
+    {
+      "epoch": 0.724200362100181,
+      "eval_entropy": 0.7380402811457601,
+      "eval_loss": 0.7054820656776428,
+      "eval_mean_token_accuracy": 0.8003534738267406,
+      "eval_num_tokens": 702638.0,
+      "eval_runtime": 88.238,
+      "eval_samples_per_second": 16.093,
+      "eval_steps_per_second": 2.017,
+      "step": 300
+    },
+    {
+      "entropy": 0.7696981988847256,
+      "epoch": 0.7724803862401931,
+      "grad_norm": 0.8408658504486084,
+      "learning_rate": 0.00028924209318837064,
+      "loss": 0.7070387363433838,
+      "mean_token_accuracy": 0.8035947173833847,
+      "num_tokens": 749377.0,
+      "step": 320
+    },
+    {
+      "epoch": 0.7724803862401931,
+      "eval_entropy": 0.7897409023193831,
+      "eval_loss": 0.7015668749809265,
+      "eval_mean_token_accuracy": 0.805306687113944,
+      "eval_num_tokens": 749377.0,
+      "eval_runtime": 87.9395,
+      "eval_samples_per_second": 16.147,
+      "eval_steps_per_second": 2.024,
+      "step": 320
+    },
+    {
+      "entropy": 0.762337189912796,
+      "epoch": 0.8207604103802052,
+      "grad_norm": 0.9843310117721558,
+      "learning_rate": 0.0003073763937017481,
+      "loss": 0.7021088123321533,
+      "mean_token_accuracy": 0.8029616877436638,
+      "num_tokens": 798874.0,
+      "step": 340
+    },
+    {
+      "epoch": 0.8207604103802052,
+      "eval_entropy": 0.7210077121016685,
+      "eval_loss": 0.7082746624946594,
+      "eval_mean_token_accuracy": 0.8049283824609906,
+      "eval_num_tokens": 798874.0,
+      "eval_runtime": 88.1706,
+      "eval_samples_per_second": 16.105,
+      "eval_steps_per_second": 2.019,
+      "step": 340
+    },
+    {
+      "entropy": 0.7879025422036647,
+      "epoch": 0.8690404345202173,
+      "grad_norm": 0.9713814854621887,
+      "learning_rate": 0.0003255106942151256,
+      "loss": 0.714624547958374,
+      "mean_token_accuracy": 0.79987031519413,
+      "num_tokens": 840825.0,
+      "step": 360
+    },
+    {
+      "epoch": 0.8690404345202173,
+      "eval_entropy": 0.7866930489459735,
+      "eval_loss": 0.7145519852638245,
+      "eval_mean_token_accuracy": 0.8015657988157165,
+      "eval_num_tokens": 840825.0,
+      "eval_runtime": 88.3746,
+      "eval_samples_per_second": 16.068,
+      "eval_steps_per_second": 2.014,
+      "step": 360
+    },
+    {
+      "entropy": 0.7844726584851742,
+      "epoch": 0.9173204586602294,
+      "grad_norm": 1.2028311491012573,
+      "learning_rate": 0.00034364499472850306,
+      "loss": 0.7040786266326904,
+      "mean_token_accuracy": 0.7981885217130185,
+      "num_tokens": 882885.0,
+      "step": 380
+    },
+    {
+      "epoch": 0.9173204586602294,
+      "eval_entropy": 0.7767668339643585,
+      "eval_loss": 0.7073574066162109,
+      "eval_mean_token_accuracy": 0.8040851833445303,
+      "eval_num_tokens": 882885.0,
+      "eval_runtime": 88.3828,
+      "eval_samples_per_second": 16.066,
+      "eval_steps_per_second": 2.014,
+      "step": 380
+    },
+    {
+      "entropy": 0.7699790760874748,
+      "epoch": 0.9656004828002414,
+      "grad_norm": 0.9397398233413696,
+      "learning_rate": 0.00036177929524188054,
+      "loss": 0.7040313720703125,
+      "mean_token_accuracy": 0.8034727744758129,
+      "num_tokens": 927456.0,
+      "step": 400
+    },
+    {
+      "epoch": 0.9656004828002414,
+      "eval_entropy": 0.7749103545472863,
+      "eval_loss": 0.7093513011932373,
+      "eval_mean_token_accuracy": 0.8019482901926791,
+      "eval_num_tokens": 927456.0,
+      "eval_runtime": 86.2112,
+      "eval_samples_per_second": 16.471,
+      "eval_steps_per_second": 2.065,
+      "step": 400
+    },
+    {
+      "entropy": 0.7672389172888422,
+      "epoch": 1.012070006035003,
+      "grad_norm": 0.99210524559021,
+      "learning_rate": 0.00037628567078152296,
+      "loss": 0.6913118362426758,
+      "mean_token_accuracy": 0.8042210944287189,
+      "num_tokens": 972662.0,
+      "step": 420
+    },
+    {
+      "epoch": 1.012070006035003,
+      "eval_entropy": 0.7294771229283193,
+      "eval_loss": 0.7171520590782166,
+      "eval_mean_token_accuracy": 0.8015128585059991,
+      "eval_num_tokens": 972662.0,
+      "eval_runtime": 88.037,
+      "eval_samples_per_second": 16.13,
+      "eval_steps_per_second": 2.022,
+      "step": 420
+    },
+    {
+      "entropy": 0.7143943183124065,
+      "epoch": 1.060350030175015,
+      "grad_norm": 1.087638258934021,
+      "learning_rate": 0.0003762484015601069,
+      "loss": 0.642704439163208,
+      "mean_token_accuracy": 0.8124966286122799,
+      "num_tokens": 1022525.0,
+      "step": 440
+    },
+    {
+      "epoch": 1.060350030175015,
+      "eval_entropy": 0.743511886744017,
+      "eval_loss": 0.7116662859916687,
+      "eval_mean_token_accuracy": 0.8037210672758939,
+      "eval_num_tokens": 1022525.0,
+      "eval_runtime": 87.9931,
+      "eval_samples_per_second": 16.138,
+      "eval_steps_per_second": 2.023,
+      "step": 440
+    },
+    {
+      "entropy": 0.7237346574664116,
+      "epoch": 1.1086300543150271,
+      "grad_norm": 0.9365553259849548,
+      "learning_rate": 0.0003761579008392123,
+      "loss": 0.652644681930542,
+      "mean_token_accuracy": 0.8107496216893196,
+      "num_tokens": 1069194.0,
+      "step": 460
+    },
+    {
+      "epoch": 1.1086300543150271,
+      "eval_entropy": 0.7372823839776972,
+      "eval_loss": 0.7238303422927856,
+      "eval_mean_token_accuracy": 0.8026230124275336,
+      "eval_num_tokens": 1069194.0,
+      "eval_runtime": 87.6778,
+      "eval_samples_per_second": 16.196,
+      "eval_steps_per_second": 2.03,
+      "step": 460
+    },
+    {
+      "entropy": 0.7202137842774391,
+      "epoch": 1.1569100784550392,
+      "grad_norm": 0.7998443841934204,
+      "learning_rate": 0.0003760141942294726,
+      "loss": 0.6485219478607178,
+      "mean_token_accuracy": 0.8092714451253414,
+      "num_tokens": 1117570.0,
+      "step": 480
+    },
+    {
+      "epoch": 1.1569100784550392,
+      "eval_entropy": 0.71547647942318,
+      "eval_loss": 0.7124277949333191,
+      "eval_mean_token_accuracy": 0.8013174219077892,
+      "eval_num_tokens": 1117570.0,
+      "eval_runtime": 87.4387,
+      "eval_samples_per_second": 16.24,
+      "eval_steps_per_second": 2.036,
+      "step": 480
+    },
+    {
+      "entropy": 0.7270819500088692,
+      "epoch": 1.2051901025950513,
+      "grad_norm": 0.9470787644386292,
+      "learning_rate": 0.00037581732239815854,
+      "loss": 0.66352219581604,
+      "mean_token_accuracy": 0.8102593503892421,
+      "num_tokens": 1161717.0,
+      "step": 500
+    },
+    {
+      "epoch": 1.2051901025950513,
+      "eval_entropy": 0.7791565897759427,
+      "eval_loss": 0.7179726958274841,
+      "eval_mean_token_accuracy": 0.7948898328154275,
+      "eval_num_tokens": 1161717.0,
+      "eval_runtime": 87.9927,
+      "eval_samples_per_second": 16.138,
+      "eval_steps_per_second": 2.023,
+      "step": 500
+    },
+    {
+      "entropy": 0.7320936664938926,
+      "epoch": 1.2534701267350634,
+      "grad_norm": 1.0220379829406738,
+      "learning_rate": 0.0003755673410576695,
+      "loss": 0.6581549644470215,
+      "mean_token_accuracy": 0.8098709337413311,
+      "num_tokens": 1209390.0,
+      "step": 520
+    },
+    {
+      "epoch": 1.2534701267350634,
+      "eval_entropy": 0.7710678225822662,
+      "eval_loss": 0.708797812461853,
+      "eval_mean_token_accuracy": 0.8039572778042783,
+      "eval_num_tokens": 1209390.0,
+      "eval_runtime": 88.1193,
+      "eval_samples_per_second": 16.115,
+      "eval_steps_per_second": 2.02,
+      "step": 520
+    },
+    {
+      "entropy": 0.7281674664467573,
+      "epoch": 1.3017501508750755,
+      "grad_norm": 0.8775188326835632,
+      "learning_rate": 0.000375264320949768,
+      "loss": 0.6631278991699219,
+      "mean_token_accuracy": 0.8101261422038079,
+      "num_tokens": 1256611.0,
+      "step": 540
+    },
+    {
+      "epoch": 1.3017501508750755,
+      "eval_entropy": 0.7199799800187014,
+      "eval_loss": 0.6981531381607056,
+      "eval_mean_token_accuracy": 0.8071265254127845,
+      "eval_num_tokens": 1256611.0,
+      "eval_runtime": 88.165,
+      "eval_samples_per_second": 16.106,
+      "eval_steps_per_second": 2.019,
+      "step": 540
+    },
+    {
+      "entropy": 0.722860773652792,
+      "epoch": 1.3500301750150876,
+      "grad_norm": 0.8949288129806519,
+      "learning_rate": 0.00037490834782556,
+      "loss": 0.6639598846435547,
+      "mean_token_accuracy": 0.8105649061501026,
+      "num_tokens": 1302325.0,
+      "step": 560
+    },
+    {
+      "epoch": 1.3500301750150876,
+      "eval_entropy": 0.7134760032209118,
+      "eval_loss": 0.7029948234558105,
+      "eval_mean_token_accuracy": 0.8062985880991046,
+      "eval_num_tokens": 1302325.0,
+      "eval_runtime": 88.1286,
+      "eval_samples_per_second": 16.113,
+      "eval_steps_per_second": 2.02,
+      "step": 560
+    },
+    {
+      "entropy": 0.7299126699566841,
+      "epoch": 1.3983101991550995,
+      "grad_norm": 1.0323039293289185,
+      "learning_rate": 0.0003744995224212291,
+      "loss": 0.6535763263702392,
+      "mean_token_accuracy": 0.8120224848389626,
+      "num_tokens": 1343844.0,
+      "step": 580
+    },
+    {
+      "epoch": 1.3983101991550995,
+      "eval_entropy": 0.7169701182440426,
+      "eval_loss": 0.7067857980728149,
+      "eval_mean_token_accuracy": 0.8040432936689826,
+      "eval_num_tokens": 1343844.0,
+      "eval_runtime": 88.2252,
+      "eval_samples_per_second": 16.095,
+      "eval_steps_per_second": 2.018,
+      "step": 580
+    },
+    {
+      "entropy": 0.7281742256134749,
+      "epoch": 1.4465902232951118,
+      "grad_norm": 1.3370424509048462,
+      "learning_rate": 0.00037403796042952863,
+      "loss": 0.6637272357940673,
+      "mean_token_accuracy": 0.8118261776864528,
+      "num_tokens": 1389471.0,
+      "step": 600
+    },
+    {
+      "epoch": 1.4465902232951118,
+      "eval_entropy": 0.7378843212395572,
+      "eval_loss": 0.7038390040397644,
+      "eval_mean_token_accuracy": 0.8059229371922739,
+      "eval_num_tokens": 1389471.0,
+      "eval_runtime": 87.9301,
+      "eval_samples_per_second": 16.149,
+      "eval_steps_per_second": 2.024,
+      "step": 600
+    },
+    {
+      "entropy": 0.7388614989817143,
+      "epoch": 1.4948702474351236,
+      "grad_norm": 0.8464171290397644,
+      "learning_rate": 0.00037352379246704257,
+      "loss": 0.6604740142822265,
+      "mean_token_accuracy": 0.8102974124252796,
+      "num_tokens": 1435975.0,
+      "step": 620
+    },
+    {
+      "epoch": 1.4948702474351236,
+      "eval_entropy": 0.7126847046814607,
+      "eval_loss": 0.6950703263282776,
+      "eval_mean_token_accuracy": 0.8085885392815879,
+      "eval_num_tokens": 1435975.0,
+      "eval_runtime": 88.2732,
+      "eval_samples_per_second": 16.086,
+      "eval_steps_per_second": 2.016,
+      "step": 620
+    },
+    {
+      "entropy": 0.7258916139602661,
+      "epoch": 1.5431502715751357,
+      "grad_norm": 1.1780085563659668,
+      "learning_rate": 0.0003729571640372223,
+      "loss": 0.6529891014099121,
+      "mean_token_accuracy": 0.8085981778800487,
+      "num_tokens": 1482035.0,
+      "step": 640
+    },
+    {
+      "epoch": 1.5431502715751357,
+      "eval_entropy": 0.7096509638797032,
+      "eval_loss": 0.690051257610321,
+      "eval_mean_token_accuracy": 0.8100520241796301,
+      "eval_num_tokens": 1482035.0,
+      "eval_runtime": 88.3153,
+      "eval_samples_per_second": 16.079,
+      "eval_steps_per_second": 2.016,
+      "step": 640
+    },
+    {
+      "entropy": 0.7151173129677773,
+      "epoch": 1.5914302957151478,
+      "grad_norm": 1.0357667207717896,
+      "learning_rate": 0.0003723382354892108,
+      "loss": 0.6451474189758301,
+      "mean_token_accuracy": 0.8136902332305909,
+      "num_tokens": 1528119.0,
+      "step": 660
+    },
+    {
+      "epoch": 1.5914302957151478,
+      "eval_entropy": 0.7060252652409371,
+      "eval_loss": 0.6899483799934387,
+      "eval_mean_token_accuracy": 0.810834194502134,
+      "eval_num_tokens": 1528119.0,
+      "eval_runtime": 87.8848,
+      "eval_samples_per_second": 16.158,
+      "eval_steps_per_second": 2.025,
+      "step": 660
+    },
+    {
+      "entropy": 0.7141377851366997,
+      "epoch": 1.63971031985516,
+      "grad_norm": 0.8391667008399963,
+      "learning_rate": 0.000371667181972466,
+      "loss": 0.6481579780578614,
+      "mean_token_accuracy": 0.8159015104174614,
+      "num_tokens": 1573661.0,
+      "step": 680
+    },
+    {
+      "epoch": 1.63971031985516,
+      "eval_entropy": 0.7598993319473909,
+      "eval_loss": 0.681204080581665,
+      "eval_mean_token_accuracy": 0.8100194586126992,
+      "eval_num_tokens": 1573661.0,
+      "eval_runtime": 88.2415,
+      "eval_samples_per_second": 16.092,
+      "eval_steps_per_second": 2.017,
+      "step": 680
+    },
+    {
+      "entropy": 0.7235980287194252,
+      "epoch": 1.687990343995172,
+      "grad_norm": 0.8197196125984192,
+      "learning_rate": 0.00037094419338719537,
+      "loss": 0.6582849025726318,
+      "mean_token_accuracy": 0.8114834539592266,
+      "num_tokens": 1621714.0,
+      "step": 700
+    },
+    {
+      "epoch": 1.687990343995172,
+      "eval_entropy": 0.7166854568411795,
+      "eval_loss": 0.679177463054657,
+      "eval_mean_token_accuracy": 0.8119039877077167,
+      "eval_num_tokens": 1621714.0,
+      "eval_runtime": 88.0957,
+      "eval_samples_per_second": 16.119,
+      "eval_steps_per_second": 2.021,
+      "step": 700
+    },
+    {
+      "entropy": 0.7278190158307553,
+      "epoch": 1.736270368135184,
+      "grad_norm": 0.7820602655410767,
+      "learning_rate": 0.0003701694743306164,
+      "loss": 0.6633153438568116,
+      "mean_token_accuracy": 0.8123249113559723,
+      "num_tokens": 1672118.0,
+      "step": 720
+    },
+    {
+      "epoch": 1.736270368135184,
+      "eval_entropy": 0.6980850314826108,
+      "eval_loss": 0.6775653958320618,
+      "eval_mean_token_accuracy": 0.8116872906684875,
+      "eval_num_tokens": 1672118.0,
+      "eval_runtime": 87.9795,
+      "eval_samples_per_second": 16.14,
+      "eval_steps_per_second": 2.023,
+      "step": 720
+    },
+    {
+      "entropy": 0.7149631556123495,
+      "epoch": 1.7845503922751962,
+      "grad_norm": 1.0139760971069336,
+      "learning_rate": 0.00036934324403905795,
+      "loss": 0.6468976020812989,
+      "mean_token_accuracy": 0.8117561548948288,
+      "num_tokens": 1720465.0,
+      "step": 740
+    },
+    {
+      "epoch": 1.7845503922751962,
+      "eval_entropy": 0.69009849094273,
+      "eval_loss": 0.6731093525886536,
+      "eval_mean_token_accuracy": 0.81281964196248,
+      "eval_num_tokens": 1720465.0,
+      "eval_runtime": 87.7412,
+      "eval_samples_per_second": 16.184,
+      "eval_steps_per_second": 2.029,
+      "step": 740
+    },
+    {
+      "entropy": 0.7226049453020096,
+      "epoch": 1.832830416415208,
+      "grad_norm": 1.5040546655654907,
+      "learning_rate": 0.0003684657363259193,
+      "loss": 0.6385328769683838,
+      "mean_token_accuracy": 0.8132639616727829,
+      "num_tokens": 1764440.0,
+      "step": 760
+    },
+    {
+      "epoch": 1.832830416415208,
+      "eval_entropy": 0.7125169618076153,
+      "eval_loss": 0.6864680647850037,
+      "eval_mean_token_accuracy": 0.8049418089095126,
+      "eval_num_tokens": 1764440.0,
+      "eval_runtime": 87.9429,
+      "eval_samples_per_second": 16.147,
+      "eval_steps_per_second": 2.024,
+      "step": 760
+    },
+    {
+      "entropy": 0.707821810990572,
+      "epoch": 1.8811104405552204,
+      "grad_norm": 0.8604223132133484,
+      "learning_rate": 0.00036753719951550327,
+      "loss": 0.6397049427032471,
+      "mean_token_accuracy": 0.8137877315282822,
+      "num_tokens": 1810828.0,
+      "step": 780
+    },
+    {
+      "epoch": 1.8811104405552204,
+      "eval_entropy": 0.6492102551326323,
+      "eval_loss": 0.6743726134300232,
+      "eval_mean_token_accuracy": 0.8130356572317273,
+      "eval_num_tokens": 1810828.0,
+      "eval_runtime": 87.8684,
+      "eval_samples_per_second": 16.161,
+      "eval_steps_per_second": 2.026,
+      "step": 780
+    },
+    {
+      "entropy": 0.6800355084240437,
+      "epoch": 1.9293904646952322,
+      "grad_norm": 0.7270982265472412,
+      "learning_rate": 0.00036655789637274377,
+      "loss": 0.6267755031585693,
+      "mean_token_accuracy": 0.8180945307016373,
+      "num_tokens": 1857600.0,
+      "step": 800
+    },
+    {
+      "epoch": 1.9293904646952322,
+      "eval_entropy": 0.6727538651294922,
+      "eval_loss": 0.6696028709411621,
+      "eval_mean_token_accuracy": 0.8116443090224534,
+      "eval_num_tokens": 1857600.0,
+      "eval_runtime": 88.2449,
+      "eval_samples_per_second": 16.092,
+      "eval_steps_per_second": 2.017,
+      "step": 800
+    },
+    {
+      "entropy": 0.6746691606938839,
+      "epoch": 1.9776704888352445,
+      "grad_norm": 0.9771540760993958,
+      "learning_rate": 0.0003655281040288461,
+      "loss": 0.6326992988586426,
+      "mean_token_accuracy": 0.8179976396262646,
+      "num_tokens": 1901998.0,
+      "step": 820
+    },
+    {
+      "epoch": 1.9776704888352445,
+      "eval_entropy": 0.6636319716324967,
+      "eval_loss": 0.6714363098144531,
+      "eval_mean_token_accuracy": 0.8126635943235976,
+      "eval_num_tokens": 1901998.0,
+      "eval_runtime": 87.7666,
+      "eval_samples_per_second": 16.179,
+      "eval_steps_per_second": 2.028,
+      "step": 820
+    },
+    {
+      "entropy": 0.6319125778489298,
+      "epoch": 2.024140012070006,
+      "grad_norm": 0.8380444645881653,
+      "learning_rate": 0.0003644481139028622,
+      "loss": 0.5780903816223144,
+      "mean_token_accuracy": 0.8250188401767186,
+      "num_tokens": 1947553.0,
+      "step": 840
+    },
+    {
+      "epoch": 2.024140012070006,
+      "eval_entropy": 0.6627151989535,
+      "eval_loss": 0.6774935722351074,
+      "eval_mean_token_accuracy": 0.8140510819601209,
+      "eval_num_tokens": 1947553.0,
+      "eval_runtime": 87.8287,
+      "eval_samples_per_second": 16.168,
+      "eval_steps_per_second": 2.027,
+      "step": 840
+    },
+    {
+      "entropy": 0.6059147048741579,
+      "epoch": 2.0724200362100182,
+      "grad_norm": 0.9745152592658997,
+      "learning_rate": 0.0003633182316192229,
+      "loss": 0.546702527999878,
+      "mean_token_accuracy": 0.8358186542987823,
+      "num_tokens": 1995234.0,
+      "step": 860
+    },
+    {
+      "epoch": 2.0724200362100182,
+      "eval_entropy": 0.6620710384979677,
+      "eval_loss": 0.6769291162490845,
+      "eval_mean_token_accuracy": 0.8135082835561773,
+      "eval_num_tokens": 1995234.0,
+      "eval_runtime": 88.0008,
+      "eval_samples_per_second": 16.136,
+      "eval_steps_per_second": 2.023,
+      "step": 860
+    },
+    {
+      "entropy": 0.6076117843389511,
+      "epoch": 2.12070006035003,
+      "grad_norm": 0.6914380192756653,
+      "learning_rate": 0.0003621387769212491,
+      "loss": 0.5324535846710206,
+      "mean_token_accuracy": 0.8366568259894848,
+      "num_tokens": 2040639.0,
+      "step": 880
+    },
+    {
+      "epoch": 2.12070006035003,
+      "eval_entropy": 0.6592224213514435,
+      "eval_loss": 0.670383632183075,
+      "eval_mean_token_accuracy": 0.8162150088320957,
+      "eval_num_tokens": 2040639.0,
+      "eval_runtime": 88.324,
+      "eval_samples_per_second": 16.077,
+      "eval_steps_per_second": 2.015,
+      "step": 880
+    },
+    {
+      "entropy": 0.6158072650432587,
+      "epoch": 2.1689800844900424,
+      "grad_norm": 0.9649198055267334,
+      "learning_rate": 0.0003609100835806688,
+      "loss": 0.5564568996429443,
+      "mean_token_accuracy": 0.8329654954373836,
+      "num_tokens": 2090604.0,
+      "step": 900
+    },
+    {
+      "epoch": 2.1689800844900424,
+      "eval_entropy": 0.608588819758276,
+      "eval_loss": 0.6745102405548096,
+      "eval_mean_token_accuracy": 0.8155080542135774,
+      "eval_num_tokens": 2090604.0,
+      "eval_runtime": 87.52,
+      "eval_samples_per_second": 16.225,
+      "eval_steps_per_second": 2.034,
+      "step": 900
+    },
+    {
+      "entropy": 0.5938734326511621,
+      "epoch": 2.2172601086300543,
+      "grad_norm": 0.6915557384490967,
+      "learning_rate": 0.0003596324993031632,
+      "loss": 0.5351875782012939,
+      "mean_token_accuracy": 0.8385631121695042,
+      "num_tokens": 2142246.0,
+      "step": 920
+    },
+    {
+      "epoch": 2.2172601086300543,
+      "eval_entropy": 0.6251631205001574,
+      "eval_loss": 0.6664518117904663,
+      "eval_mean_token_accuracy": 0.81711940417129,
+      "eval_num_tokens": 2142246.0,
+      "eval_runtime": 87.1792,
+      "eval_samples_per_second": 16.288,
+      "eval_steps_per_second": 2.042,
+      "step": 920
+    },
+    {
+      "entropy": 0.623173613846302,
+      "epoch": 2.2655401327700666,
+      "grad_norm": 0.8366610407829285,
+      "learning_rate": 0.00035830638562997063,
+      "loss": 0.5488213539123535,
+      "mean_token_accuracy": 0.8321483485400677,
+      "num_tokens": 2183144.0,
+      "step": 940
+    },
+    {
+      "epoch": 2.2655401327700666,
+      "eval_entropy": 0.6237085649136747,
+      "eval_loss": 0.68075031042099,
+      "eval_mean_token_accuracy": 0.814671738429016,
+      "eval_num_tokens": 2183144.0,
+      "eval_runtime": 88.0652,
+      "eval_samples_per_second": 16.124,
+      "eval_steps_per_second": 2.021,
+      "step": 940
+    },
+    {
+      "entropy": 0.6236911740154027,
+      "epoch": 2.3138201569100785,
+      "grad_norm": 0.7812452912330627,
+      "learning_rate": 0.00035693211783557416,
+      "loss": 0.5696200847625732,
+      "mean_token_accuracy": 0.82991396561265,
+      "num_tokens": 2228098.0,
+      "step": 960
+    },
+    {
+      "epoch": 2.3138201569100785,
+      "eval_entropy": 0.6327033426319615,
+      "eval_loss": 0.6742984652519226,
+      "eval_mean_token_accuracy": 0.8148086091105857,
+      "eval_num_tokens": 2228098.0,
+      "eval_runtime": 88.1711,
+      "eval_samples_per_second": 16.105,
+      "eval_steps_per_second": 2.019,
+      "step": 960
+    },
+    {
+      "entropy": 0.6218426413834095,
+      "epoch": 2.3621001810500903,
+      "grad_norm": 0.8719790577888489,
+      "learning_rate": 0.0003555100848215035,
+      "loss": 0.5582651615142822,
+      "mean_token_accuracy": 0.8323961742222309,
+      "num_tokens": 2273062.0,
+      "step": 980
+    },
+    {
+      "epoch": 2.3621001810500903,
+      "eval_entropy": 0.6338012375858393,
+      "eval_loss": 0.6685822606086731,
+      "eval_mean_token_accuracy": 0.8152180300669724,
+      "eval_num_tokens": 2273062.0,
+      "eval_runtime": 88.101,
+      "eval_samples_per_second": 16.118,
+      "eval_steps_per_second": 2.02,
+      "step": 980
+    },
+    {
+      "entropy": 0.6050681818276644,
+      "epoch": 2.4103802051901027,
+      "grad_norm": 0.9151566028594971,
+      "learning_rate": 0.00035404068900628076,
+      "loss": 0.559452486038208,
+      "mean_token_accuracy": 0.8310242861509323,
+      "num_tokens": 2319305.0,
+      "step": 1000
+    },
+    {
+      "epoch": 2.4103802051901027,
+      "eval_entropy": 0.6634890969549672,
+      "eval_loss": 0.6692465543746948,
+      "eval_mean_token_accuracy": 0.8134353673190213,
+      "eval_num_tokens": 2319305.0,
+      "eval_runtime": 87.7711,
+      "eval_samples_per_second": 16.178,
+      "eval_steps_per_second": 2.028,
+      "step": 1000
+    }
+  ],
+  "logging_steps": 20,
+  "max_steps": 4150,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 10,
+  "save_steps": 20,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 9.8038608775637e+16,
+  "train_batch_size": 4,
+  "trial_name": null,
+  "trial_params": null
+}

overgeneralisation_original_Estonian/Qwen3.5-4B-Base_overgeneralisation_splits_original_features_train_overgeneralisation_splits_original_features_test2/checkpoint-1020/README.md

@@ -0,0 +1,209 @@
+---
+base_model: Qwen/Qwen3.5-4B-Base
+library_name: peft
+pipeline_tag: text-generation
+tags:
+- base_model:adapter:Qwen/Qwen3.5-4B-Base
+- lora
+- sft
+- transformers
+- trl
+---
+
+# Model Card for Model ID
+
+<!-- Provide a quick summary of what the model is/does. -->
+
+
+
+## Model Details
+
+### Model Description
+
+<!-- Provide a longer summary of what this model is. -->
+
+
+
+- **Developed by:** [More Information Needed]
+- **Funded by [optional]:** [More Information Needed]
+- **Shared by [optional]:** [More Information Needed]
+- **Model type:** [More Information Needed]
+- **Language(s) (NLP):** [More Information Needed]
+- **License:** [More Information Needed]
+- **Finetuned from model [optional]:** [More Information Needed]
+
+### Model Sources [optional]
+
+<!-- Provide the basic links for the model. -->
+
+- **Repository:** [More Information Needed]
+- **Paper [optional]:** [More Information Needed]
+- **Demo [optional]:** [More Information Needed]
+
+## Uses
+
+<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
+
+### Direct Use
+
+<!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
+
+[More Information Needed]
+
+### Downstream Use [optional]
+
+<!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->
+
+[More Information Needed]
+
+### Out-of-Scope Use
+
+<!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
+
+[More Information Needed]
+
+## Bias, Risks, and Limitations
+
+<!-- This section is meant to convey both technical and sociotechnical limitations. -->
+
+[More Information Needed]
+
+### Recommendations
+
+<!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->
+
+Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
+
+## How to Get Started with the Model
+
+Use the code below to get started with the model.
+
+[More Information Needed]
+
+## Training Details
+
+### Training Data
+
+<!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
+
+[More Information Needed]
+
+### Training Procedure
+
+<!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
+
+#### Preprocessing [optional]
+
+[More Information Needed]
+
+
+#### Training Hyperparameters
+
+- **Training regime:** [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->
+
+#### Speeds, Sizes, Times [optional]
+
+<!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
+
+[More Information Needed]
+
+## Evaluation
+
+<!-- This section describes the evaluation protocols and provides the results. -->
+
+### Testing Data, Factors & Metrics
+
+#### Testing Data
+
+<!-- This should link to a Dataset Card if possible. -->
+
+[More Information Needed]
+
+#### Factors
+
+<!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->
+
+[More Information Needed]
+
+#### Metrics
+
+<!-- These are the evaluation metrics being used, ideally with a description of why. -->
+
+[More Information Needed]
+
+### Results
+
+[More Information Needed]
+
+#### Summary
+
+
+
+## Model Examination [optional]
+
+<!-- Relevant interpretability work for the model goes here -->
+
+[More Information Needed]
+
+## Environmental Impact
+
+<!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly -->
+
+Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700).
+
+- **Hardware Type:** [More Information Needed]
+- **Hours used:** [More Information Needed]
+- **Cloud Provider:** [More Information Needed]
+- **Compute Region:** [More Information Needed]
+- **Carbon Emitted:** [More Information Needed]
+
+## Technical Specifications [optional]
+
+### Model Architecture and Objective
+
+[More Information Needed]
+
+### Compute Infrastructure
+
+[More Information Needed]
+
+#### Hardware
+
+[More Information Needed]
+
+#### Software
+
+[More Information Needed]
+
+## Citation [optional]
+
+<!-- If there is a paper or blog post introducing the model, the APA and Bibtex information for that should go in this section. -->
+
+**BibTeX:**
+
+[More Information Needed]
+
+**APA:**
+
+[More Information Needed]
+
+## Glossary [optional]
+
+<!-- If relevant, include terms and calculations in this section that can help readers understand the model or model card. -->
+
+[More Information Needed]
+
+## More Information [optional]
+
+[More Information Needed]
+
+## Model Card Authors [optional]
+
+[More Information Needed]
+
+## Model Card Contact
+
+[More Information Needed]
+### Framework versions
+
+- PEFT 0.18.1

overgeneralisation_original_Estonian/Qwen3.5-4B-Base_overgeneralisation_splits_original_features_train_overgeneralisation_splits_original_features_test2/checkpoint-1020/adapter_config.json

@@ -0,0 +1,46 @@
+{
+  "alora_invocation_tokens": null,
+  "alpha_pattern": {},
+  "arrow_config": null,
+  "auto_mapping": null,
+  "base_model_name_or_path": "Qwen/Qwen3.5-4B-Base",
+  "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": 128,
+  "lora_bias": false,
+  "lora_dropout": 0.03828026524568501,
+  "megatron_config": null,
+  "megatron_core": "megatron.core",
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "peft_version": "0.18.1",
+  "qalora_group_size": 16,
+  "r": 64,
+  "rank_pattern": {},
+  "revision": null,
+  "target_modules": [
+    "gate_proj",
+    "v_proj",
+    "up_proj",
+    "down_proj",
+    "o_proj",
+    "q_proj",
+    "k_proj"
+  ],
+  "target_parameters": null,
+  "task_type": "CAUSAL_LM",
+  "trainable_token_indices": null,
+  "use_dora": false,
+  "use_qalora": false,
+  "use_rslora": false
+}

overgeneralisation_original_Estonian/Qwen3.5-4B-Base_overgeneralisation_splits_original_features_train_overgeneralisation_splits_original_features_test2/checkpoint-1020/chat_template.jinja

@@ -0,0 +1,154 @@
+{%- set image_count = namespace(value=0) %}
+{%- set video_count = namespace(value=0) %}
+{%- macro render_content(content, do_vision_count, is_system_content=false) %}
+    {%- if content is string %}
+        {{- content }}
+    {%- elif content is iterable and content is not mapping %}
+        {%- for item in content %}
+            {%- if 'image' in item or 'image_url' in item or item.type == 'image' %}
+                {%- if is_system_content %}
+                    {{- raise_exception('System message cannot contain images.') }}
+                {%- endif %}
+                {%- if do_vision_count %}
+                    {%- set image_count.value = image_count.value + 1 %}
+                {%- endif %}
+                {%- if add_vision_id %}
+                    {{- 'Picture ' ~ image_count.value ~ ': ' }}
+                {%- endif %}
+                {{- '<|vision_start|><|image_pad|><|vision_end|>' }}
+            {%- elif 'video' in item or item.type == 'video' %}
+                {%- if is_system_content %}
+                    {{- raise_exception('System message cannot contain videos.') }}
+                {%- endif %}
+                {%- if do_vision_count %}
+                    {%- set video_count.value = video_count.value + 1 %}
+                {%- endif %}
+                {%- if add_vision_id %}
+                    {{- 'Video ' ~ video_count.value ~ ': ' }}
+                {%- endif %}
+                {{- '<|vision_start|><|video_pad|><|vision_end|>' }}
+            {%- elif 'text' in item %}
+                {{- item.text }}
+            {%- else %}
+                {{- raise_exception('Unexpected item type in content.') }}
+            {%- endif %}
+        {%- endfor %}
+    {%- elif content is none or content is undefined %}
+        {{- '' }}
+    {%- else %}
+        {{- raise_exception('Unexpected content type.') }}
+    {%- endif %}
+{%- endmacro %}
+{%- if not messages %}
+    {{- raise_exception('No messages provided.') }}
+{%- endif %}
+{%- if tools and tools is iterable and tools is not mapping %}
+    {{- '<|im_start|>system\n' }}
+    {{- "# Tools\n\nYou have access to the following functions:\n\n<tools>" }}
+    {%- for tool in tools %}
+        {{- "\n" }}
+        {{- tool | tojson }}
+    {%- endfor %}
+    {{- "\n</tools>" }}
+    {{- '\n\nIf you choose to call a function ONLY reply in the following format with NO suffix:\n\n<tool_call>\n<function=example_function_name>\n<parameter=example_parameter_1>\nvalue_1\n</parameter>\n<parameter=example_parameter_2>\nThis is the value for the second parameter\nthat can span\nmultiple lines\n</parameter>\n</function>\n</tool_call>\n\n<IMPORTANT>\nReminder:\n- Function calls MUST follow the specified format: an inner <function=...></function> block must be nested within <tool_call></tool_call> XML tags\n- Required parameters MUST be specified\n- You may provide optional reasoning for your function call in natural language BEFORE the function call, but NOT after\n- If there is no function call available, answer the question like normal with your current knowledge and do not tell the user about function calls\n</IMPORTANT>' }}
+    {%- if messages[0].role == 'system' %}
+        {%- set content = render_content(messages[0].content, false, true)|trim %}
+        {%- if content %}
+            {{- '\n\n' + content }}
+        {%- endif %}
+    {%- endif %}
+    {{- '<|im_end|>\n' }}
+{%- else %}
+    {%- if messages[0].role == 'system' %}
+        {%- set content = render_content(messages[0].content, false, true)|trim %}
+        {{- '<|im_start|>system\n' + content + '<|im_end|>\n' }}
+    {%- endif %}
+{%- endif %}
+{%- set ns = namespace(multi_step_tool=true, last_query_index=messages|length - 1) %}
+{%- for message in messages[::-1] %}
+    {%- set index = (messages|length - 1) - loop.index0 %}
+    {%- if ns.multi_step_tool and message.role == "user" %}
+        {%- set content = render_content(message.content, false)|trim %}
+        {%- if not(content.startswith('<tool_response>') and content.endswith('</tool_response>')) %}
+            {%- set ns.multi_step_tool = false %}
+            {%- set ns.last_query_index = index %}
+        {%- endif %}
+    {%- endif %}
+{%- endfor %}
+{%- if ns.multi_step_tool %}
+    {{- raise_exception('No user query found in messages.') }}
+{%- endif %}
+{%- for message in messages %}
+    {%- set content = render_content(message.content, true)|trim %}
+    {%- if message.role == "system" %}
+        {%- if not loop.first %}
+            {{- raise_exception('System message must be at the beginning.') }}
+        {%- endif %}
+    {%- elif message.role == "user" %}
+        {{- '<|im_start|>' + message.role + '\n' + content + '<|im_end|>' + '\n' }}
+    {%- elif message.role == "assistant" %}
+        {%- set reasoning_content = '' %}
+        {%- if message.reasoning_content is string %}
+            {%- set reasoning_content = message.reasoning_content %}
+        {%- else %}
+            {%- if '</think>' in content %}
+                {%- set reasoning_content = content.split('</think>')[0].rstrip('\n').split('<think>')[-1].lstrip('\n') %}
+                {%- set content = content.split('</think>')[-1].lstrip('\n') %}
+            {%- endif %}
+        {%- endif %}
+        {%- set reasoning_content = reasoning_content|trim %}
+        {%- if loop.index0 > ns.last_query_index %}
+            {{- '<|im_start|>' + message.role + '\n<think>\n' + reasoning_content + '\n</think>\n\n' + content }}
+        {%- else %}
+            {{- '<|im_start|>' + message.role + '\n' + content }}
+        {%- endif %}
+        {%- if message.tool_calls and message.tool_calls is iterable and message.tool_calls is not mapping %}
+            {%- for tool_call in message.tool_calls %}
+                {%- if tool_call.function is defined %}
+                    {%- set tool_call = tool_call.function %}
+                {%- endif %}
+                {%- if loop.first %}
+                    {%- if content|trim %}
+                        {{- '\n\n<tool_call>\n<function=' + tool_call.name + '>\n' }}
+                    {%- else %}
+                        {{- '<tool_call>\n<function=' + tool_call.name + '>\n' }}
+                    {%- endif %}
+                {%- else %}
+                    {{- '\n<tool_call>\n<function=' + tool_call.name + '>\n' }}
+                {%- endif %}
+                {%- if tool_call.arguments is defined %}
+                    {%- for args_name, args_value in tool_call.arguments|items %}
+                        {{- '<parameter=' + args_name + '>\n' }}
+                        {%- set args_value = args_value | tojson | safe if args_value is mapping or (args_value is sequence and args_value is not string) else args_value | string %}
+                        {{- args_value }}
+                        {{- '\n</parameter>\n' }}
+                    {%- endfor %}
+                {%- endif %}
+                {{- '</function>\n</tool_call>' }}
+            {%- endfor %}
+        {%- endif %}
+        {{- '<|im_end|>\n' }}
+    {%- elif message.role == "tool" %}
+        {%- if loop.previtem and loop.previtem.role != "tool" %}
+            {{- '<|im_start|>user' }}
+        {%- endif %}
+        {{- '\n<tool_response>\n' }}
+        {{- content }}
+        {{- '\n</tool_response>' }}
+        {%- if not loop.last and loop.nextitem.role != "tool" %}
+            {{- '<|im_end|>\n' }}
+        {%- elif loop.last %}
+            {{- '<|im_end|>\n' }}
+        {%- endif %}
+    {%- else %}
+        {{- raise_exception('Unexpected message role.') }}
+    {%- endif %}
+{%- endfor %}
+{%- if add_generation_prompt %}
+    {{- '<|im_start|>assistant\n' }}
+    {%- if enable_thinking is defined and enable_thinking is false %}
+        {{- '<think>\n\n</think>\n\n' }}
+    {%- else %}
+        {{- '<think>\n' }}
+    {%- endif %}
+{%- endif %}

overgeneralisation_original_Estonian/Qwen3.5-4B-Base_overgeneralisation_splits_original_features_train_overgeneralisation_splits_original_features_test2/checkpoint-1020/tokenizer_config.json

@@ -0,0 +1,31 @@
+{
+  "add_prefix_space": false,
+  "audio_bos_token": "<|audio_start|>",
+  "audio_eos_token": "<|audio_end|>",
+  "audio_token": "<|audio_pad|>",
+  "backend": "tokenizers",
+  "bos_token": null,
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<|endoftext|>",
+  "errors": "replace",
+  "image_token": "<|image_pad|>",
+  "is_local": false,
+  "model_max_length": 262144,
+  "model_specific_special_tokens": {
+    "audio_bos_token": "<|audio_start|>",
+    "audio_eos_token": "<|audio_end|>",
+    "audio_token": "<|audio_pad|>",
+    "image_token": "<|image_pad|>",
+    "video_token": "<|video_pad|>",
+    "vision_bos_token": "<|vision_start|>",
+    "vision_eos_token": "<|vision_end|>"
+  },
+  "pad_token": "<|endoftext|>",
+  "pretokenize_regex": "(?i:'s|'t|'re|'ve|'m|'ll|'d)|[^\\r\\n\\p{L}\\p{N}]?[\\p{L}\\p{M}]+|\\p{N}| ?[^\\s\\p{L}\\p{M}\\p{N}]+[\\r\\n]*|\\s*[\\r\\n]+|\\s+(?!\\S)|\\s+",
+  "split_special_tokens": false,
+  "tokenizer_class": "TokenizersBackend",
+  "unk_token": null,
+  "video_token": "<|video_pad|>",
+  "vision_bos_token": "<|vision_start|>",
+  "vision_eos_token": "<|vision_end|>"
+}

overgeneralisation_original_Estonian/Qwen3.5-4B-Base_overgeneralisation_splits_original_features_train_overgeneralisation_splits_original_features_test2/checkpoint-1020/trainer_state.json

@@ -0,0 +1,1105 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 2.4586602293301145,
+  "eval_steps": 20,
+  "global_step": 1020,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "entropy": 2.172999547421932,
+      "epoch": 0.04828002414001207,
+      "grad_norm": 2.5071208477020264,
+      "learning_rate": 1.7227585487708594e-05,
+      "loss": 1.9347333908081055,
+      "mean_token_accuracy": 0.6086816020309925,
+      "num_tokens": 47778.0,
+      "step": 20
+    },
+    {
+      "epoch": 0.04828002414001207,
+      "eval_entropy": 1.6909754791956269,
+      "eval_loss": 1.4728385210037231,
+      "eval_mean_token_accuracy": 0.6773593797442619,
+      "eval_num_tokens": 47778.0,
+      "eval_runtime": 90.1265,
+      "eval_samples_per_second": 15.756,
+      "eval_steps_per_second": 1.975,
+      "step": 20
+    },
+    {
+      "entropy": 1.2014626950025558,
+      "epoch": 0.09656004828002414,
+      "grad_norm": 1.5074543952941895,
+      "learning_rate": 3.5361886001086065e-05,
+      "loss": 1.0867146492004394,
+      "mean_token_accuracy": 0.7285927519202232,
+      "num_tokens": 95981.0,
+      "step": 40
+    },
+    {
+      "epoch": 0.09656004828002414,
+      "eval_entropy": 1.0304168352250302,
+      "eval_loss": 0.9211422204971313,
+      "eval_mean_token_accuracy": 0.7613122848312507,
+      "eval_num_tokens": 95981.0,
+      "eval_runtime": 88.6338,
+      "eval_samples_per_second": 16.021,
+      "eval_steps_per_second": 2.008,
+      "step": 40
+    },
+    {
+      "entropy": 0.9454198732972146,
+      "epoch": 0.14484007242003621,
+      "grad_norm": 1.7578108310699463,
+      "learning_rate": 5.349618651446354e-05,
+      "loss": 0.8447297096252442,
+      "mean_token_accuracy": 0.7723424732685089,
+      "num_tokens": 142784.0,
+      "step": 60
+    },
+    {
+      "epoch": 0.14484007242003621,
+      "eval_entropy": 0.8647834474451086,
+      "eval_loss": 0.8240737318992615,
+      "eval_mean_token_accuracy": 0.7821227014064789,
+      "eval_num_tokens": 142784.0,
+      "eval_runtime": 88.0135,
+      "eval_samples_per_second": 16.134,
+      "eval_steps_per_second": 2.022,
+      "step": 60
+    },
+    {
+      "entropy": 0.8963550612330436,
+      "epoch": 0.19312009656004828,
+      "grad_norm": 1.1464931964874268,
+      "learning_rate": 7.1630487027841e-05,
+      "loss": 0.7979836463928223,
+      "mean_token_accuracy": 0.7822489373385906,
+      "num_tokens": 187338.0,
+      "step": 80
+    },
+    {
+      "epoch": 0.19312009656004828,
+      "eval_entropy": 0.8663296133614657,
+      "eval_loss": 0.7752988338470459,
+      "eval_mean_token_accuracy": 0.7885957012015782,
+      "eval_num_tokens": 187338.0,
+      "eval_runtime": 88.284,
+      "eval_samples_per_second": 16.084,
+      "eval_steps_per_second": 2.016,
+      "step": 80
+    },
+    {
+      "entropy": 0.8265103206038475,
+      "epoch": 0.24140012070006034,
+      "grad_norm": 1.0106662511825562,
+      "learning_rate": 8.976478754121848e-05,
+      "loss": 0.7297060012817382,
+      "mean_token_accuracy": 0.7955138415098191,
+      "num_tokens": 236038.0,
+      "step": 100
+    },
+    {
+      "epoch": 0.24140012070006034,
+      "eval_entropy": 0.8122169536151244,
+      "eval_loss": 0.754906177520752,
+      "eval_mean_token_accuracy": 0.7917155255092664,
+      "eval_num_tokens": 236038.0,
+      "eval_runtime": 87.5754,
+      "eval_samples_per_second": 16.215,
+      "eval_steps_per_second": 2.033,
+      "step": 100
+    },
+    {
+      "entropy": 0.8098392844200134,
+      "epoch": 0.28968014484007243,
+      "grad_norm": 1.2238041162490845,
+      "learning_rate": 0.00010789908805459595,
+      "loss": 0.7201489448547364,
+      "mean_token_accuracy": 0.7975016921758652,
+      "num_tokens": 282918.0,
+      "step": 120
+    },
+    {
+      "epoch": 0.28968014484007243,
+      "eval_entropy": 0.8418726462326692,
+      "eval_loss": 0.7393178939819336,
+      "eval_mean_token_accuracy": 0.7936771585700217,
+      "eval_num_tokens": 282918.0,
+      "eval_runtime": 87.1597,
+      "eval_samples_per_second": 16.292,
+      "eval_steps_per_second": 2.042,
+      "step": 120
+    },
+    {
+      "entropy": 0.8080591082572937,
+      "epoch": 0.33796016898008446,
+      "grad_norm": 1.011591911315918,
+      "learning_rate": 0.00012603338856797343,
+      "loss": 0.7147697925567627,
+      "mean_token_accuracy": 0.7947175532579422,
+      "num_tokens": 331305.0,
+      "step": 140
+    },
+    {
+      "epoch": 0.33796016898008446,
+      "eval_entropy": 0.7894941431083037,
+      "eval_loss": 0.720219075679779,
+      "eval_mean_token_accuracy": 0.801899236239744,
+      "eval_num_tokens": 331305.0,
+      "eval_runtime": 87.1493,
+      "eval_samples_per_second": 16.294,
+      "eval_steps_per_second": 2.042,
+      "step": 140
+    },
+    {
+      "entropy": 0.7957205109298229,
+      "epoch": 0.38624019312009655,
+      "grad_norm": 0.8481830954551697,
+      "learning_rate": 0.00014416768908135088,
+      "loss": 0.7050027847290039,
+      "mean_token_accuracy": 0.7995368830859662,
+      "num_tokens": 375876.0,
+      "step": 160
+    },
+    {
+      "epoch": 0.38624019312009655,
+      "eval_entropy": 0.7746140591883928,
+      "eval_loss": 0.7175475358963013,
+      "eval_mean_token_accuracy": 0.7994848955213354,
+      "eval_num_tokens": 375876.0,
+      "eval_runtime": 86.655,
+      "eval_samples_per_second": 16.387,
+      "eval_steps_per_second": 2.054,
+      "step": 160
+    },
+    {
+      "entropy": 0.779565304517746,
+      "epoch": 0.43452021726010864,
+      "grad_norm": 1.032624363899231,
+      "learning_rate": 0.00016230198959472835,
+      "loss": 0.6978332042694092,
+      "mean_token_accuracy": 0.8015618294477462,
+      "num_tokens": 423913.0,
+      "step": 180
+    },
+    {
+      "epoch": 0.43452021726010864,
+      "eval_entropy": 0.7455583871080634,
+      "eval_loss": 0.7118850350379944,
+      "eval_mean_token_accuracy": 0.7999682195400923,
+      "eval_num_tokens": 423913.0,
+      "eval_runtime": 86.7842,
+      "eval_samples_per_second": 16.362,
+      "eval_steps_per_second": 2.051,
+      "step": 180
+    },
+    {
+      "entropy": 0.7810172818601131,
+      "epoch": 0.4828002414001207,
+      "grad_norm": 0.9276663064956665,
+      "learning_rate": 0.00018043629010810582,
+      "loss": 0.6937861442565918,
+      "mean_token_accuracy": 0.802893053740263,
+      "num_tokens": 472395.0,
+      "step": 200
+    },
+    {
+      "epoch": 0.4828002414001207,
+      "eval_entropy": 0.7650193290764027,
+      "eval_loss": 0.7020567059516907,
+      "eval_mean_token_accuracy": 0.8045302153973097,
+      "eval_num_tokens": 472395.0,
+      "eval_runtime": 88.326,
+      "eval_samples_per_second": 16.077,
+      "eval_steps_per_second": 2.015,
+      "step": 200
+    },
+    {
+      "entropy": 0.7862060204148292,
+      "epoch": 0.5310802655401328,
+      "grad_norm": 0.8195295929908752,
+      "learning_rate": 0.0001985705906214833,
+      "loss": 0.6878782272338867,
+      "mean_token_accuracy": 0.8027168907225132,
+      "num_tokens": 517049.0,
+      "step": 220
+    },
+    {
+      "epoch": 0.5310802655401328,
+      "eval_entropy": 0.7656565583154057,
+      "eval_loss": 0.6973932981491089,
+      "eval_mean_token_accuracy": 0.8047967278555538,
+      "eval_num_tokens": 517049.0,
+      "eval_runtime": 87.7698,
+      "eval_samples_per_second": 16.179,
+      "eval_steps_per_second": 2.028,
+      "step": 220
+    },
+    {
+      "entropy": 0.7611534461379051,
+      "epoch": 0.5793602896801449,
+      "grad_norm": 0.7499712109565735,
+      "learning_rate": 0.00021670489113486077,
+      "loss": 0.681937837600708,
+      "mean_token_accuracy": 0.8041978515684605,
+      "num_tokens": 562836.0,
+      "step": 240
+    },
+    {
+      "epoch": 0.5793602896801449,
+      "eval_entropy": 0.7599442380197933,
+      "eval_loss": 0.7002046704292297,
+      "eval_mean_token_accuracy": 0.8051861517884759,
+      "eval_num_tokens": 562836.0,
+      "eval_runtime": 87.9239,
+      "eval_samples_per_second": 16.15,
+      "eval_steps_per_second": 2.024,
+      "step": 240
+    },
+    {
+      "entropy": 0.7800783462822437,
+      "epoch": 0.627640313820157,
+      "grad_norm": 0.8645269274711609,
+      "learning_rate": 0.00023483919164823825,
+      "loss": 0.6954500675201416,
+      "mean_token_accuracy": 0.8031484372913837,
+      "num_tokens": 606612.0,
+      "step": 260
+    },
+    {
+      "epoch": 0.627640313820157,
+      "eval_entropy": 0.7498679090751691,
+      "eval_loss": 0.7046149373054504,
+      "eval_mean_token_accuracy": 0.802388215667746,
+      "eval_num_tokens": 606612.0,
+      "eval_runtime": 88.2468,
+      "eval_samples_per_second": 16.091,
+      "eval_steps_per_second": 2.017,
+      "step": 260
+    },
+    {
+      "entropy": 0.7586533665657044,
+      "epoch": 0.6759203379601689,
+      "grad_norm": 0.741596519947052,
+      "learning_rate": 0.0002529734921616157,
+      "loss": 0.690476369857788,
+      "mean_token_accuracy": 0.8046923100948333,
+      "num_tokens": 655920.0,
+      "step": 280
+    },
+    {
+      "epoch": 0.6759203379601689,
+      "eval_entropy": 0.7714274066887544,
+      "eval_loss": 0.6992160677909851,
+      "eval_mean_token_accuracy": 0.8042646482419432,
+      "eval_num_tokens": 655920.0,
+      "eval_runtime": 88.1754,
+      "eval_samples_per_second": 16.104,
+      "eval_steps_per_second": 2.019,
+      "step": 280
+    },
+    {
+      "entropy": 0.7585848286747933,
+      "epoch": 0.724200362100181,
+      "grad_norm": 1.0778124332427979,
+      "learning_rate": 0.00027110779267499317,
+      "loss": 0.6790409564971924,
+      "mean_token_accuracy": 0.8060351841151714,
+      "num_tokens": 702638.0,
+      "step": 300
+    },
+    {
+      "epoch": 0.724200362100181,
+      "eval_entropy": 0.7380402811457601,
+      "eval_loss": 0.7054820656776428,
+      "eval_mean_token_accuracy": 0.8003534738267406,
+      "eval_num_tokens": 702638.0,
+      "eval_runtime": 88.238,
+      "eval_samples_per_second": 16.093,
+      "eval_steps_per_second": 2.017,
+      "step": 300
+    },
+    {
+      "entropy": 0.7696981988847256,
+      "epoch": 0.7724803862401931,
+      "grad_norm": 0.8408658504486084,
+      "learning_rate": 0.00028924209318837064,
+      "loss": 0.7070387363433838,
+      "mean_token_accuracy": 0.8035947173833847,
+      "num_tokens": 749377.0,
+      "step": 320
+    },
+    {
+      "epoch": 0.7724803862401931,
+      "eval_entropy": 0.7897409023193831,
+      "eval_loss": 0.7015668749809265,
+      "eval_mean_token_accuracy": 0.805306687113944,
+      "eval_num_tokens": 749377.0,
+      "eval_runtime": 87.9395,
+      "eval_samples_per_second": 16.147,
+      "eval_steps_per_second": 2.024,
+      "step": 320
+    },
+    {
+      "entropy": 0.762337189912796,
+      "epoch": 0.8207604103802052,
+      "grad_norm": 0.9843310117721558,
+      "learning_rate": 0.0003073763937017481,
+      "loss": 0.7021088123321533,
+      "mean_token_accuracy": 0.8029616877436638,
+      "num_tokens": 798874.0,
+      "step": 340
+    },
+    {
+      "epoch": 0.8207604103802052,
+      "eval_entropy": 0.7210077121016685,
+      "eval_loss": 0.7082746624946594,
+      "eval_mean_token_accuracy": 0.8049283824609906,
+      "eval_num_tokens": 798874.0,
+      "eval_runtime": 88.1706,
+      "eval_samples_per_second": 16.105,
+      "eval_steps_per_second": 2.019,
+      "step": 340
+    },
+    {
+      "entropy": 0.7879025422036647,
+      "epoch": 0.8690404345202173,
+      "grad_norm": 0.9713814854621887,
+      "learning_rate": 0.0003255106942151256,
+      "loss": 0.714624547958374,
+      "mean_token_accuracy": 0.79987031519413,
+      "num_tokens": 840825.0,
+      "step": 360
+    },
+    {
+      "epoch": 0.8690404345202173,
+      "eval_entropy": 0.7866930489459735,
+      "eval_loss": 0.7145519852638245,
+      "eval_mean_token_accuracy": 0.8015657988157165,
+      "eval_num_tokens": 840825.0,
+      "eval_runtime": 88.3746,
+      "eval_samples_per_second": 16.068,
+      "eval_steps_per_second": 2.014,
+      "step": 360
+    },
+    {
+      "entropy": 0.7844726584851742,
+      "epoch": 0.9173204586602294,
+      "grad_norm": 1.2028311491012573,
+      "learning_rate": 0.00034364499472850306,
+      "loss": 0.7040786266326904,
+      "mean_token_accuracy": 0.7981885217130185,
+      "num_tokens": 882885.0,
+      "step": 380
+    },
+    {
+      "epoch": 0.9173204586602294,
+      "eval_entropy": 0.7767668339643585,
+      "eval_loss": 0.7073574066162109,
+      "eval_mean_token_accuracy": 0.8040851833445303,
+      "eval_num_tokens": 882885.0,
+      "eval_runtime": 88.3828,
+      "eval_samples_per_second": 16.066,
+      "eval_steps_per_second": 2.014,
+      "step": 380
+    },
+    {
+      "entropy": 0.7699790760874748,
+      "epoch": 0.9656004828002414,
+      "grad_norm": 0.9397398233413696,
+      "learning_rate": 0.00036177929524188054,
+      "loss": 0.7040313720703125,
+      "mean_token_accuracy": 0.8034727744758129,
+      "num_tokens": 927456.0,
+      "step": 400
+    },
+    {
+      "epoch": 0.9656004828002414,
+      "eval_entropy": 0.7749103545472863,
+      "eval_loss": 0.7093513011932373,
+      "eval_mean_token_accuracy": 0.8019482901926791,
+      "eval_num_tokens": 927456.0,
+      "eval_runtime": 86.2112,
+      "eval_samples_per_second": 16.471,
+      "eval_steps_per_second": 2.065,
+      "step": 400
+    },
+    {
+      "entropy": 0.7672389172888422,
+      "epoch": 1.012070006035003,
+      "grad_norm": 0.99210524559021,
+      "learning_rate": 0.00037628567078152296,
+      "loss": 0.6913118362426758,
+      "mean_token_accuracy": 0.8042210944287189,
+      "num_tokens": 972662.0,
+      "step": 420
+    },
+    {
+      "epoch": 1.012070006035003,
+      "eval_entropy": 0.7294771229283193,
+      "eval_loss": 0.7171520590782166,
+      "eval_mean_token_accuracy": 0.8015128585059991,
+      "eval_num_tokens": 972662.0,
+      "eval_runtime": 88.037,
+      "eval_samples_per_second": 16.13,
+      "eval_steps_per_second": 2.022,
+      "step": 420
+    },
+    {
+      "entropy": 0.7143943183124065,
+      "epoch": 1.060350030175015,
+      "grad_norm": 1.087638258934021,
+      "learning_rate": 0.0003762484015601069,
+      "loss": 0.642704439163208,
+      "mean_token_accuracy": 0.8124966286122799,
+      "num_tokens": 1022525.0,
+      "step": 440
+    },
+    {
+      "epoch": 1.060350030175015,
+      "eval_entropy": 0.743511886744017,
+      "eval_loss": 0.7116662859916687,
+      "eval_mean_token_accuracy": 0.8037210672758939,
+      "eval_num_tokens": 1022525.0,
+      "eval_runtime": 87.9931,
+      "eval_samples_per_second": 16.138,
+      "eval_steps_per_second": 2.023,
+      "step": 440
+    },
+    {
+      "entropy": 0.7237346574664116,
+      "epoch": 1.1086300543150271,
+      "grad_norm": 0.9365553259849548,
+      "learning_rate": 0.0003761579008392123,
+      "loss": 0.652644681930542,
+      "mean_token_accuracy": 0.8107496216893196,
+      "num_tokens": 1069194.0,
+      "step": 460
+    },
+    {
+      "epoch": 1.1086300543150271,
+      "eval_entropy": 0.7372823839776972,
+      "eval_loss": 0.7238303422927856,
+      "eval_mean_token_accuracy": 0.8026230124275336,
+      "eval_num_tokens": 1069194.0,
+      "eval_runtime": 87.6778,
+      "eval_samples_per_second": 16.196,
+      "eval_steps_per_second": 2.03,
+      "step": 460
+    },
+    {
+      "entropy": 0.7202137842774391,
+      "epoch": 1.1569100784550392,
+      "grad_norm": 0.7998443841934204,
+      "learning_rate": 0.0003760141942294726,
+      "loss": 0.6485219478607178,
+      "mean_token_accuracy": 0.8092714451253414,
+      "num_tokens": 1117570.0,
+      "step": 480
+    },
+    {
+      "epoch": 1.1569100784550392,
+      "eval_entropy": 0.71547647942318,
+      "eval_loss": 0.7124277949333191,
+      "eval_mean_token_accuracy": 0.8013174219077892,
+      "eval_num_tokens": 1117570.0,
+      "eval_runtime": 87.4387,
+      "eval_samples_per_second": 16.24,
+      "eval_steps_per_second": 2.036,
+      "step": 480
+    },
+    {
+      "entropy": 0.7270819500088692,
+      "epoch": 1.2051901025950513,
+      "grad_norm": 0.9470787644386292,
+      "learning_rate": 0.00037581732239815854,
+      "loss": 0.66352219581604,
+      "mean_token_accuracy": 0.8102593503892421,
+      "num_tokens": 1161717.0,
+      "step": 500
+    },
+    {
+      "epoch": 1.2051901025950513,
+      "eval_entropy": 0.7791565897759427,
+      "eval_loss": 0.7179726958274841,
+      "eval_mean_token_accuracy": 0.7948898328154275,
+      "eval_num_tokens": 1161717.0,
+      "eval_runtime": 87.9927,
+      "eval_samples_per_second": 16.138,
+      "eval_steps_per_second": 2.023,
+      "step": 500
+    },
+    {
+      "entropy": 0.7320936664938926,
+      "epoch": 1.2534701267350634,
+      "grad_norm": 1.0220379829406738,
+      "learning_rate": 0.0003755673410576695,
+      "loss": 0.6581549644470215,
+      "mean_token_accuracy": 0.8098709337413311,
+      "num_tokens": 1209390.0,
+      "step": 520
+    },
+    {
+      "epoch": 1.2534701267350634,
+      "eval_entropy": 0.7710678225822662,
+      "eval_loss": 0.708797812461853,
+      "eval_mean_token_accuracy": 0.8039572778042783,
+      "eval_num_tokens": 1209390.0,
+      "eval_runtime": 88.1193,
+      "eval_samples_per_second": 16.115,
+      "eval_steps_per_second": 2.02,
+      "step": 520
+    },
+    {
+      "entropy": 0.7281674664467573,
+      "epoch": 1.3017501508750755,
+      "grad_norm": 0.8775188326835632,
+      "learning_rate": 0.000375264320949768,
+      "loss": 0.6631278991699219,
+      "mean_token_accuracy": 0.8101261422038079,
+      "num_tokens": 1256611.0,
+      "step": 540
+    },
+    {
+      "epoch": 1.3017501508750755,
+      "eval_entropy": 0.7199799800187014,
+      "eval_loss": 0.6981531381607056,
+      "eval_mean_token_accuracy": 0.8071265254127845,
+      "eval_num_tokens": 1256611.0,
+      "eval_runtime": 88.165,
+      "eval_samples_per_second": 16.106,
+      "eval_steps_per_second": 2.019,
+      "step": 540
+    },
+    {
+      "entropy": 0.722860773652792,
+      "epoch": 1.3500301750150876,
+      "grad_norm": 0.8949288129806519,
+      "learning_rate": 0.00037490834782556,
+      "loss": 0.6639598846435547,
+      "mean_token_accuracy": 0.8105649061501026,
+      "num_tokens": 1302325.0,
+      "step": 560
+    },
+    {
+      "epoch": 1.3500301750150876,
+      "eval_entropy": 0.7134760032209118,
+      "eval_loss": 0.7029948234558105,
+      "eval_mean_token_accuracy": 0.8062985880991046,
+      "eval_num_tokens": 1302325.0,
+      "eval_runtime": 88.1286,
+      "eval_samples_per_second": 16.113,
+      "eval_steps_per_second": 2.02,
+      "step": 560
+    },
+    {
+      "entropy": 0.7299126699566841,
+      "epoch": 1.3983101991550995,
+      "grad_norm": 1.0323039293289185,
+      "learning_rate": 0.0003744995224212291,
+      "loss": 0.6535763263702392,
+      "mean_token_accuracy": 0.8120224848389626,
+      "num_tokens": 1343844.0,
+      "step": 580
+    },
+    {
+      "epoch": 1.3983101991550995,
+      "eval_entropy": 0.7169701182440426,
+      "eval_loss": 0.7067857980728149,
+      "eval_mean_token_accuracy": 0.8040432936689826,
+      "eval_num_tokens": 1343844.0,
+      "eval_runtime": 88.2252,
+      "eval_samples_per_second": 16.095,
+      "eval_steps_per_second": 2.018,
+      "step": 580
+    },
+    {
+      "entropy": 0.7281742256134749,
+      "epoch": 1.4465902232951118,
+      "grad_norm": 1.3370424509048462,
+      "learning_rate": 0.00037403796042952863,
+      "loss": 0.6637272357940673,
+      "mean_token_accuracy": 0.8118261776864528,
+      "num_tokens": 1389471.0,
+      "step": 600
+    },
+    {
+      "epoch": 1.4465902232951118,
+      "eval_entropy": 0.7378843212395572,
+      "eval_loss": 0.7038390040397644,
+      "eval_mean_token_accuracy": 0.8059229371922739,
+      "eval_num_tokens": 1389471.0,
+      "eval_runtime": 87.9301,
+      "eval_samples_per_second": 16.149,
+      "eval_steps_per_second": 2.024,
+      "step": 600
+    },
+    {
+      "entropy": 0.7388614989817143,
+      "epoch": 1.4948702474351236,
+      "grad_norm": 0.8464171290397644,
+      "learning_rate": 0.00037352379246704257,
+      "loss": 0.6604740142822265,
+      "mean_token_accuracy": 0.8102974124252796,
+      "num_tokens": 1435975.0,
+      "step": 620
+    },
+    {
+      "epoch": 1.4948702474351236,
+      "eval_entropy": 0.7126847046814607,
+      "eval_loss": 0.6950703263282776,
+      "eval_mean_token_accuracy": 0.8085885392815879,
+      "eval_num_tokens": 1435975.0,
+      "eval_runtime": 88.2732,
+      "eval_samples_per_second": 16.086,
+      "eval_steps_per_second": 2.016,
+      "step": 620
+    },
+    {
+      "entropy": 0.7258916139602661,
+      "epoch": 1.5431502715751357,
+      "grad_norm": 1.1780085563659668,
+      "learning_rate": 0.0003729571640372223,
+      "loss": 0.6529891014099121,
+      "mean_token_accuracy": 0.8085981778800487,
+      "num_tokens": 1482035.0,
+      "step": 640
+    },
+    {
+      "epoch": 1.5431502715751357,
+      "eval_entropy": 0.7096509638797032,
+      "eval_loss": 0.690051257610321,
+      "eval_mean_token_accuracy": 0.8100520241796301,
+      "eval_num_tokens": 1482035.0,
+      "eval_runtime": 88.3153,
+      "eval_samples_per_second": 16.079,
+      "eval_steps_per_second": 2.016,
+      "step": 640
+    },
+    {
+      "entropy": 0.7151173129677773,
+      "epoch": 1.5914302957151478,
+      "grad_norm": 1.0357667207717896,
+      "learning_rate": 0.0003723382354892108,
+      "loss": 0.6451474189758301,
+      "mean_token_accuracy": 0.8136902332305909,
+      "num_tokens": 1528119.0,
+      "step": 660
+    },
+    {
+      "epoch": 1.5914302957151478,
+      "eval_entropy": 0.7060252652409371,
+      "eval_loss": 0.6899483799934387,
+      "eval_mean_token_accuracy": 0.810834194502134,
+      "eval_num_tokens": 1528119.0,
+      "eval_runtime": 87.8848,
+      "eval_samples_per_second": 16.158,
+      "eval_steps_per_second": 2.025,
+      "step": 660
+    },
+    {
+      "entropy": 0.7141377851366997,
+      "epoch": 1.63971031985516,
+      "grad_norm": 0.8391667008399963,
+      "learning_rate": 0.000371667181972466,
+      "loss": 0.6481579780578614,
+      "mean_token_accuracy": 0.8159015104174614,
+      "num_tokens": 1573661.0,
+      "step": 680
+    },
+    {
+      "epoch": 1.63971031985516,
+      "eval_entropy": 0.7598993319473909,
+      "eval_loss": 0.681204080581665,
+      "eval_mean_token_accuracy": 0.8100194586126992,
+      "eval_num_tokens": 1573661.0,
+      "eval_runtime": 88.2415,
+      "eval_samples_per_second": 16.092,
+      "eval_steps_per_second": 2.017,
+      "step": 680
+    },
+    {
+      "entropy": 0.7235980287194252,
+      "epoch": 1.687990343995172,
+      "grad_norm": 0.8197196125984192,
+      "learning_rate": 0.00037094419338719537,
+      "loss": 0.6582849025726318,
+      "mean_token_accuracy": 0.8114834539592266,
+      "num_tokens": 1621714.0,
+      "step": 700
+    },
+    {
+      "epoch": 1.687990343995172,
+      "eval_entropy": 0.7166854568411795,
+      "eval_loss": 0.679177463054657,
+      "eval_mean_token_accuracy": 0.8119039877077167,
+      "eval_num_tokens": 1621714.0,
+      "eval_runtime": 88.0957,
+      "eval_samples_per_second": 16.119,
+      "eval_steps_per_second": 2.021,
+      "step": 700
+    },
+    {
+      "entropy": 0.7278190158307553,
+      "epoch": 1.736270368135184,
+      "grad_norm": 0.7820602655410767,
+      "learning_rate": 0.0003701694743306164,
+      "loss": 0.6633153438568116,
+      "mean_token_accuracy": 0.8123249113559723,
+      "num_tokens": 1672118.0,
+      "step": 720
+    },
+    {
+      "epoch": 1.736270368135184,
+      "eval_entropy": 0.6980850314826108,
+      "eval_loss": 0.6775653958320618,
+      "eval_mean_token_accuracy": 0.8116872906684875,
+      "eval_num_tokens": 1672118.0,
+      "eval_runtime": 87.9795,
+      "eval_samples_per_second": 16.14,
+      "eval_steps_per_second": 2.023,
+      "step": 720
+    },
+    {
+      "entropy": 0.7149631556123495,
+      "epoch": 1.7845503922751962,
+      "grad_norm": 1.0139760971069336,
+      "learning_rate": 0.00036934324403905795,
+      "loss": 0.6468976020812989,
+      "mean_token_accuracy": 0.8117561548948288,
+      "num_tokens": 1720465.0,
+      "step": 740
+    },
+    {
+      "epoch": 1.7845503922751962,
+      "eval_entropy": 0.69009849094273,
+      "eval_loss": 0.6731093525886536,
+      "eval_mean_token_accuracy": 0.81281964196248,
+      "eval_num_tokens": 1720465.0,
+      "eval_runtime": 87.7412,
+      "eval_samples_per_second": 16.184,
+      "eval_steps_per_second": 2.029,
+      "step": 740
+    },
+    {
+      "entropy": 0.7226049453020096,
+      "epoch": 1.832830416415208,
+      "grad_norm": 1.5040546655654907,
+      "learning_rate": 0.0003684657363259193,
+      "loss": 0.6385328769683838,
+      "mean_token_accuracy": 0.8132639616727829,
+      "num_tokens": 1764440.0,
+      "step": 760
+    },
+    {
+      "epoch": 1.832830416415208,
+      "eval_entropy": 0.7125169618076153,
+      "eval_loss": 0.6864680647850037,
+      "eval_mean_token_accuracy": 0.8049418089095126,
+      "eval_num_tokens": 1764440.0,
+      "eval_runtime": 87.9429,
+      "eval_samples_per_second": 16.147,
+      "eval_steps_per_second": 2.024,
+      "step": 760
+    },
+    {
+      "entropy": 0.707821810990572,
+      "epoch": 1.8811104405552204,
+      "grad_norm": 0.8604223132133484,
+      "learning_rate": 0.00036753719951550327,
+      "loss": 0.6397049427032471,
+      "mean_token_accuracy": 0.8137877315282822,
+      "num_tokens": 1810828.0,
+      "step": 780
+    },
+    {
+      "epoch": 1.8811104405552204,
+      "eval_entropy": 0.6492102551326323,
+      "eval_loss": 0.6743726134300232,
+      "eval_mean_token_accuracy": 0.8130356572317273,
+      "eval_num_tokens": 1810828.0,
+      "eval_runtime": 87.8684,
+      "eval_samples_per_second": 16.161,
+      "eval_steps_per_second": 2.026,
+      "step": 780
+    },
+    {
+      "entropy": 0.6800355084240437,
+      "epoch": 1.9293904646952322,
+      "grad_norm": 0.7270982265472412,
+      "learning_rate": 0.00036655789637274377,
+      "loss": 0.6267755031585693,
+      "mean_token_accuracy": 0.8180945307016373,
+      "num_tokens": 1857600.0,
+      "step": 800
+    },
+    {
+      "epoch": 1.9293904646952322,
+      "eval_entropy": 0.6727538651294922,
+      "eval_loss": 0.6696028709411621,
+      "eval_mean_token_accuracy": 0.8116443090224534,
+      "eval_num_tokens": 1857600.0,
+      "eval_runtime": 88.2449,
+      "eval_samples_per_second": 16.092,
+      "eval_steps_per_second": 2.017,
+      "step": 800
+    },
+    {
+      "entropy": 0.6746691606938839,
+      "epoch": 1.9776704888352445,
+      "grad_norm": 0.9771540760993958,
+      "learning_rate": 0.0003655281040288461,
+      "loss": 0.6326992988586426,
+      "mean_token_accuracy": 0.8179976396262646,
+      "num_tokens": 1901998.0,
+      "step": 820
+    },
+    {
+      "epoch": 1.9776704888352445,
+      "eval_entropy": 0.6636319716324967,
+      "eval_loss": 0.6714363098144531,
+      "eval_mean_token_accuracy": 0.8126635943235976,
+      "eval_num_tokens": 1901998.0,
+      "eval_runtime": 87.7666,
+      "eval_samples_per_second": 16.179,
+      "eval_steps_per_second": 2.028,
+      "step": 820
+    },
+    {
+      "entropy": 0.6319125778489298,
+      "epoch": 2.024140012070006,
+      "grad_norm": 0.8380444645881653,
+      "learning_rate": 0.0003644481139028622,
+      "loss": 0.5780903816223144,
+      "mean_token_accuracy": 0.8250188401767186,
+      "num_tokens": 1947553.0,
+      "step": 840
+    },
+    {
+      "epoch": 2.024140012070006,
+      "eval_entropy": 0.6627151989535,
+      "eval_loss": 0.6774935722351074,
+      "eval_mean_token_accuracy": 0.8140510819601209,
+      "eval_num_tokens": 1947553.0,
+      "eval_runtime": 87.8287,
+      "eval_samples_per_second": 16.168,
+      "eval_steps_per_second": 2.027,
+      "step": 840
+    },
+    {
+      "entropy": 0.6059147048741579,
+      "epoch": 2.0724200362100182,
+      "grad_norm": 0.9745152592658997,
+      "learning_rate": 0.0003633182316192229,
+      "loss": 0.546702527999878,
+      "mean_token_accuracy": 0.8358186542987823,
+      "num_tokens": 1995234.0,
+      "step": 860
+    },
+    {
+      "epoch": 2.0724200362100182,
+      "eval_entropy": 0.6620710384979677,
+      "eval_loss": 0.6769291162490845,
+      "eval_mean_token_accuracy": 0.8135082835561773,
+      "eval_num_tokens": 1995234.0,
+      "eval_runtime": 88.0008,
+      "eval_samples_per_second": 16.136,
+      "eval_steps_per_second": 2.023,
+      "step": 860
+    },
+    {
+      "entropy": 0.6076117843389511,
+      "epoch": 2.12070006035003,
+      "grad_norm": 0.6914380192756653,
+      "learning_rate": 0.0003621387769212491,
+      "loss": 0.5324535846710206,
+      "mean_token_accuracy": 0.8366568259894848,
+      "num_tokens": 2040639.0,
+      "step": 880
+    },
+    {
+      "epoch": 2.12070006035003,
+      "eval_entropy": 0.6592224213514435,
+      "eval_loss": 0.670383632183075,
+      "eval_mean_token_accuracy": 0.8162150088320957,
+      "eval_num_tokens": 2040639.0,
+      "eval_runtime": 88.324,
+      "eval_samples_per_second": 16.077,
+      "eval_steps_per_second": 2.015,
+      "step": 880
+    },
+    {
+      "entropy": 0.6158072650432587,
+      "epoch": 2.1689800844900424,
+      "grad_norm": 0.9649198055267334,
+      "learning_rate": 0.0003609100835806688,
+      "loss": 0.5564568996429443,
+      "mean_token_accuracy": 0.8329654954373836,
+      "num_tokens": 2090604.0,
+      "step": 900
+    },
+    {
+      "epoch": 2.1689800844900424,
+      "eval_entropy": 0.608588819758276,
+      "eval_loss": 0.6745102405548096,
+      "eval_mean_token_accuracy": 0.8155080542135774,
+      "eval_num_tokens": 2090604.0,
+      "eval_runtime": 87.52,
+      "eval_samples_per_second": 16.225,
+      "eval_steps_per_second": 2.034,
+      "step": 900
+    },
+    {
+      "entropy": 0.5938734326511621,
+      "epoch": 2.2172601086300543,
+      "grad_norm": 0.6915557384490967,
+      "learning_rate": 0.0003596324993031632,
+      "loss": 0.5351875782012939,
+      "mean_token_accuracy": 0.8385631121695042,
+      "num_tokens": 2142246.0,
+      "step": 920
+    },
+    {
+      "epoch": 2.2172601086300543,
+      "eval_entropy": 0.6251631205001574,
+      "eval_loss": 0.6664518117904663,
+      "eval_mean_token_accuracy": 0.81711940417129,
+      "eval_num_tokens": 2142246.0,
+      "eval_runtime": 87.1792,
+      "eval_samples_per_second": 16.288,
+      "eval_steps_per_second": 2.042,
+      "step": 920
+    },
+    {
+      "entropy": 0.623173613846302,
+      "epoch": 2.2655401327700666,
+      "grad_norm": 0.8366610407829285,
+      "learning_rate": 0.00035830638562997063,
+      "loss": 0.5488213539123535,
+      "mean_token_accuracy": 0.8321483485400677,
+      "num_tokens": 2183144.0,
+      "step": 940
+    },
+    {
+      "epoch": 2.2655401327700666,
+      "eval_entropy": 0.6237085649136747,
+      "eval_loss": 0.68075031042099,
+      "eval_mean_token_accuracy": 0.814671738429016,
+      "eval_num_tokens": 2183144.0,
+      "eval_runtime": 88.0652,
+      "eval_samples_per_second": 16.124,
+      "eval_steps_per_second": 2.021,
+      "step": 940
+    },
+    {
+      "entropy": 0.6236911740154027,
+      "epoch": 2.3138201569100785,
+      "grad_norm": 0.7812452912330627,
+      "learning_rate": 0.00035693211783557416,
+      "loss": 0.5696200847625732,
+      "mean_token_accuracy": 0.82991396561265,
+      "num_tokens": 2228098.0,
+      "step": 960
+    },
+    {
+      "epoch": 2.3138201569100785,
+      "eval_entropy": 0.6327033426319615,
+      "eval_loss": 0.6742984652519226,
+      "eval_mean_token_accuracy": 0.8148086091105857,
+      "eval_num_tokens": 2228098.0,
+      "eval_runtime": 88.1711,
+      "eval_samples_per_second": 16.105,
+      "eval_steps_per_second": 2.019,
+      "step": 960
+    },
+    {
+      "entropy": 0.6218426413834095,
+      "epoch": 2.3621001810500903,
+      "grad_norm": 0.8719790577888489,
+      "learning_rate": 0.0003555100848215035,
+      "loss": 0.5582651615142822,
+      "mean_token_accuracy": 0.8323961742222309,
+      "num_tokens": 2273062.0,
+      "step": 980
+    },
+    {
+      "epoch": 2.3621001810500903,
+      "eval_entropy": 0.6338012375858393,
+      "eval_loss": 0.6685822606086731,
+      "eval_mean_token_accuracy": 0.8152180300669724,
+      "eval_num_tokens": 2273062.0,
+      "eval_runtime": 88.101,
+      "eval_samples_per_second": 16.118,
+      "eval_steps_per_second": 2.02,
+      "step": 980
+    },
+    {
+      "entropy": 0.6050681818276644,
+      "epoch": 2.4103802051901027,
+      "grad_norm": 0.9151566028594971,
+      "learning_rate": 0.00035404068900628076,
+      "loss": 0.559452486038208,
+      "mean_token_accuracy": 0.8310242861509323,
+      "num_tokens": 2319305.0,
+      "step": 1000
+    },
+    {
+      "epoch": 2.4103802051901027,
+      "eval_entropy": 0.6634890969549672,
+      "eval_loss": 0.6692465543746948,
+      "eval_mean_token_accuracy": 0.8134353673190213,
+      "eval_num_tokens": 2319305.0,
+      "eval_runtime": 87.7711,
+      "eval_samples_per_second": 16.178,
+      "eval_steps_per_second": 2.028,
+      "step": 1000
+    },
+    {
+      "entropy": 0.6169968567788601,
+      "epoch": 2.4586602293301145,
+      "grad_norm": 0.7081006765365601,
+      "learning_rate": 0.0003525243462115402,
+      "loss": 0.5589694023132324,
+      "mean_token_accuracy": 0.8357502184808254,
+      "num_tokens": 2368592.0,
+      "step": 1020
+    },
+    {
+      "epoch": 2.4586602293301145,
+      "eval_entropy": 0.6126393578695447,
+      "eval_loss": 0.6706712245941162,
+      "eval_mean_token_accuracy": 0.8155577500884452,
+      "eval_num_tokens": 2368592.0,
+      "eval_runtime": 88.1407,
+      "eval_samples_per_second": 16.111,
+      "eval_steps_per_second": 2.019,
+      "step": 1020
+    }
+  ],
+  "logging_steps": 20,
+  "max_steps": 4150,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 10,
+  "save_steps": 20,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 1.0007168784646349e+17,
+  "train_batch_size": 4,
+  "trial_name": null,
+  "trial_params": null
+}

overgeneralisation_original_Estonian/Qwen3.5-4B-Base_overgeneralisation_splits_original_features_train_overgeneralisation_splits_original_features_test2/checkpoint-1040/README.md

@@ -0,0 +1,209 @@
+---
+base_model: Qwen/Qwen3.5-4B-Base
+library_name: peft
+pipeline_tag: text-generation
+tags:
+- base_model:adapter:Qwen/Qwen3.5-4B-Base
+- lora
+- sft
+- transformers
+- trl
+---
+
+# Model Card for Model ID
+
+<!-- Provide a quick summary of what the model is/does. -->
+
+
+
+## Model Details
+
+### Model Description
+
+<!-- Provide a longer summary of what this model is. -->
+
+
+
+- **Developed by:** [More Information Needed]
+- **Funded by [optional]:** [More Information Needed]
+- **Shared by [optional]:** [More Information Needed]
+- **Model type:** [More Information Needed]
+- **Language(s) (NLP):** [More Information Needed]
+- **License:** [More Information Needed]
+- **Finetuned from model [optional]:** [More Information Needed]
+
+### Model Sources [optional]
+
+<!-- Provide the basic links for the model. -->
+
+- **Repository:** [More Information Needed]
+- **Paper [optional]:** [More Information Needed]
+- **Demo [optional]:** [More Information Needed]
+
+## Uses
+
+<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
+
+### Direct Use
+
+<!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
+
+[More Information Needed]
+
+### Downstream Use [optional]
+
+<!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->
+
+[More Information Needed]
+
+### Out-of-Scope Use
+
+<!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
+
+[More Information Needed]
+
+## Bias, Risks, and Limitations
+
+<!-- This section is meant to convey both technical and sociotechnical limitations. -->
+
+[More Information Needed]
+
+### Recommendations
+
+<!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->
+
+Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
+
+## How to Get Started with the Model
+
+Use the code below to get started with the model.
+
+[More Information Needed]
+
+## Training Details
+
+### Training Data
+
+<!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
+
+[More Information Needed]
+
+### Training Procedure
+
+<!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
+
+#### Preprocessing [optional]
+
+[More Information Needed]
+
+
+#### Training Hyperparameters
+
+- **Training regime:** [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->
+
+#### Speeds, Sizes, Times [optional]
+
+<!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
+
+[More Information Needed]
+
+## Evaluation
+
+<!-- This section describes the evaluation protocols and provides the results. -->
+
+### Testing Data, Factors & Metrics
+
+#### Testing Data
+
+<!-- This should link to a Dataset Card if possible. -->
+
+[More Information Needed]
+
+#### Factors
+
+<!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->
+
+[More Information Needed]
+
+#### Metrics
+
+<!-- These are the evaluation metrics being used, ideally with a description of why. -->
+
+[More Information Needed]
+
+### Results
+
+[More Information Needed]
+
+#### Summary
+
+
+
+## Model Examination [optional]
+
+<!-- Relevant interpretability work for the model goes here -->
+
+[More Information Needed]
+
+## Environmental Impact
+
+<!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly -->
+
+Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700).
+
+- **Hardware Type:** [More Information Needed]
+- **Hours used:** [More Information Needed]
+- **Cloud Provider:** [More Information Needed]
+- **Compute Region:** [More Information Needed]
+- **Carbon Emitted:** [More Information Needed]
+
+## Technical Specifications [optional]
+
+### Model Architecture and Objective
+
+[More Information Needed]
+
+### Compute Infrastructure
+
+[More Information Needed]
+
+#### Hardware
+
+[More Information Needed]
+
+#### Software
+
+[More Information Needed]
+
+## Citation [optional]
+
+<!-- If there is a paper or blog post introducing the model, the APA and Bibtex information for that should go in this section. -->
+
+**BibTeX:**
+
+[More Information Needed]
+
+**APA:**
+
+[More Information Needed]
+
+## Glossary [optional]
+
+<!-- If relevant, include terms and calculations in this section that can help readers understand the model or model card. -->
+
+[More Information Needed]
+
+## More Information [optional]
+
+[More Information Needed]
+
+## Model Card Authors [optional]
+
+[More Information Needed]
+
+## Model Card Contact
+
+[More Information Needed]
+### Framework versions
+
+- PEFT 0.18.1

overgeneralisation_original_Estonian/Qwen3.5-4B-Base_overgeneralisation_splits_original_features_train_overgeneralisation_splits_original_features_test2/checkpoint-1040/adapter_config.json

@@ -0,0 +1,46 @@
+{
+  "alora_invocation_tokens": null,
+  "alpha_pattern": {},
+  "arrow_config": null,
+  "auto_mapping": null,
+  "base_model_name_or_path": "Qwen/Qwen3.5-4B-Base",
+  "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": 128,
+  "lora_bias": false,
+  "lora_dropout": 0.03828026524568501,
+  "megatron_config": null,
+  "megatron_core": "megatron.core",
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "peft_version": "0.18.1",
+  "qalora_group_size": 16,
+  "r": 64,
+  "rank_pattern": {},
+  "revision": null,
+  "target_modules": [
+    "gate_proj",
+    "v_proj",
+    "up_proj",
+    "down_proj",
+    "o_proj",
+    "q_proj",
+    "k_proj"
+  ],
+  "target_parameters": null,
+  "task_type": "CAUSAL_LM",
+  "trainable_token_indices": null,
+  "use_dora": false,
+  "use_qalora": false,
+  "use_rslora": false
+}

overgeneralisation_original_Estonian/Qwen3.5-4B-Base_overgeneralisation_splits_original_features_train_overgeneralisation_splits_original_features_test2/checkpoint-1040/chat_template.jinja

@@ -0,0 +1,154 @@
+{%- set image_count = namespace(value=0) %}
+{%- set video_count = namespace(value=0) %}
+{%- macro render_content(content, do_vision_count, is_system_content=false) %}
+    {%- if content is string %}
+        {{- content }}
+    {%- elif content is iterable and content is not mapping %}
+        {%- for item in content %}
+            {%- if 'image' in item or 'image_url' in item or item.type == 'image' %}
+                {%- if is_system_content %}
+                    {{- raise_exception('System message cannot contain images.') }}
+                {%- endif %}
+                {%- if do_vision_count %}
+                    {%- set image_count.value = image_count.value + 1 %}
+                {%- endif %}
+                {%- if add_vision_id %}
+                    {{- 'Picture ' ~ image_count.value ~ ': ' }}
+                {%- endif %}
+                {{- '<|vision_start|><|image_pad|><|vision_end|>' }}
+            {%- elif 'video' in item or item.type == 'video' %}
+                {%- if is_system_content %}
+                    {{- raise_exception('System message cannot contain videos.') }}
+                {%- endif %}
+                {%- if do_vision_count %}
+                    {%- set video_count.value = video_count.value + 1 %}
+                {%- endif %}
+                {%- if add_vision_id %}
+                    {{- 'Video ' ~ video_count.value ~ ': ' }}
+                {%- endif %}
+                {{- '<|vision_start|><|video_pad|><|vision_end|>' }}
+            {%- elif 'text' in item %}
+                {{- item.text }}
+            {%- else %}
+                {{- raise_exception('Unexpected item type in content.') }}
+            {%- endif %}
+        {%- endfor %}
+    {%- elif content is none or content is undefined %}
+        {{- '' }}
+    {%- else %}
+        {{- raise_exception('Unexpected content type.') }}
+    {%- endif %}
+{%- endmacro %}
+{%- if not messages %}
+    {{- raise_exception('No messages provided.') }}
+{%- endif %}
+{%- if tools and tools is iterable and tools is not mapping %}
+    {{- '<|im_start|>system\n' }}
+    {{- "# Tools\n\nYou have access to the following functions:\n\n<tools>" }}
+    {%- for tool in tools %}
+        {{- "\n" }}
+        {{- tool | tojson }}
+    {%- endfor %}
+    {{- "\n</tools>" }}
+    {{- '\n\nIf you choose to call a function ONLY reply in the following format with NO suffix:\n\n<tool_call>\n<function=example_function_name>\n<parameter=example_parameter_1>\nvalue_1\n</parameter>\n<parameter=example_parameter_2>\nThis is the value for the second parameter\nthat can span\nmultiple lines\n</parameter>\n</function>\n</tool_call>\n\n<IMPORTANT>\nReminder:\n- Function calls MUST follow the specified format: an inner <function=...></function> block must be nested within <tool_call></tool_call> XML tags\n- Required parameters MUST be specified\n- You may provide optional reasoning for your function call in natural language BEFORE the function call, but NOT after\n- If there is no function call available, answer the question like normal with your current knowledge and do not tell the user about function calls\n</IMPORTANT>' }}
+    {%- if messages[0].role == 'system' %}
+        {%- set content = render_content(messages[0].content, false, true)|trim %}
+        {%- if content %}
+            {{- '\n\n' + content }}
+        {%- endif %}
+    {%- endif %}
+    {{- '<|im_end|>\n' }}
+{%- else %}
+    {%- if messages[0].role == 'system' %}
+        {%- set content = render_content(messages[0].content, false, true)|trim %}
+        {{- '<|im_start|>system\n' + content + '<|im_end|>\n' }}
+    {%- endif %}
+{%- endif %}
+{%- set ns = namespace(multi_step_tool=true, last_query_index=messages|length - 1) %}
+{%- for message in messages[::-1] %}
+    {%- set index = (messages|length - 1) - loop.index0 %}
+    {%- if ns.multi_step_tool and message.role == "user" %}
+        {%- set content = render_content(message.content, false)|trim %}
+        {%- if not(content.startswith('<tool_response>') and content.endswith('</tool_response>')) %}
+            {%- set ns.multi_step_tool = false %}
+            {%- set ns.last_query_index = index %}
+        {%- endif %}
+    {%- endif %}
+{%- endfor %}
+{%- if ns.multi_step_tool %}
+    {{- raise_exception('No user query found in messages.') }}
+{%- endif %}
+{%- for message in messages %}
+    {%- set content = render_content(message.content, true)|trim %}
+    {%- if message.role == "system" %}
+        {%- if not loop.first %}
+            {{- raise_exception('System message must be at the beginning.') }}
+        {%- endif %}
+    {%- elif message.role == "user" %}
+        {{- '<|im_start|>' + message.role + '\n' + content + '<|im_end|>' + '\n' }}
+    {%- elif message.role == "assistant" %}
+        {%- set reasoning_content = '' %}
+        {%- if message.reasoning_content is string %}
+            {%- set reasoning_content = message.reasoning_content %}
+        {%- else %}
+            {%- if '</think>' in content %}
+                {%- set reasoning_content = content.split('</think>')[0].rstrip('\n').split('<think>')[-1].lstrip('\n') %}
+                {%- set content = content.split('</think>')[-1].lstrip('\n') %}
+            {%- endif %}
+        {%- endif %}
+        {%- set reasoning_content = reasoning_content|trim %}
+        {%- if loop.index0 > ns.last_query_index %}
+            {{- '<|im_start|>' + message.role + '\n<think>\n' + reasoning_content + '\n</think>\n\n' + content }}
+        {%- else %}
+            {{- '<|im_start|>' + message.role + '\n' + content }}
+        {%- endif %}
+        {%- if message.tool_calls and message.tool_calls is iterable and message.tool_calls is not mapping %}
+            {%- for tool_call in message.tool_calls %}
+                {%- if tool_call.function is defined %}
+                    {%- set tool_call = tool_call.function %}
+                {%- endif %}
+                {%- if loop.first %}
+                    {%- if content|trim %}
+                        {{- '\n\n<tool_call>\n<function=' + tool_call.name + '>\n' }}
+                    {%- else %}
+                        {{- '<tool_call>\n<function=' + tool_call.name + '>\n' }}
+                    {%- endif %}
+                {%- else %}
+                    {{- '\n<tool_call>\n<function=' + tool_call.name + '>\n' }}
+                {%- endif %}
+                {%- if tool_call.arguments is defined %}
+                    {%- for args_name, args_value in tool_call.arguments|items %}
+                        {{- '<parameter=' + args_name + '>\n' }}
+                        {%- set args_value = args_value | tojson | safe if args_value is mapping or (args_value is sequence and args_value is not string) else args_value | string %}
+                        {{- args_value }}
+                        {{- '\n</parameter>\n' }}
+                    {%- endfor %}
+                {%- endif %}
+                {{- '</function>\n</tool_call>' }}
+            {%- endfor %}
+        {%- endif %}
+        {{- '<|im_end|>\n' }}
+    {%- elif message.role == "tool" %}
+        {%- if loop.previtem and loop.previtem.role != "tool" %}
+            {{- '<|im_start|>user' }}
+        {%- endif %}
+        {{- '\n<tool_response>\n' }}
+        {{- content }}
+        {{- '\n</tool_response>' }}
+        {%- if not loop.last and loop.nextitem.role != "tool" %}
+            {{- '<|im_end|>\n' }}
+        {%- elif loop.last %}
+            {{- '<|im_end|>\n' }}
+        {%- endif %}
+    {%- else %}
+        {{- raise_exception('Unexpected message role.') }}
+    {%- endif %}
+{%- endfor %}
+{%- if add_generation_prompt %}
+    {{- '<|im_start|>assistant\n' }}
+    {%- if enable_thinking is defined and enable_thinking is false %}
+        {{- '<think>\n\n</think>\n\n' }}
+    {%- else %}
+        {{- '<think>\n' }}
+    {%- endif %}
+{%- endif %}

overgeneralisation_original_Estonian/Qwen3.5-4B-Base_overgeneralisation_splits_original_features_train_overgeneralisation_splits_original_features_test2/checkpoint-1040/tokenizer_config.json

@@ -0,0 +1,31 @@
+{
+  "add_prefix_space": false,
+  "audio_bos_token": "<|audio_start|>",
+  "audio_eos_token": "<|audio_end|>",
+  "audio_token": "<|audio_pad|>",
+  "backend": "tokenizers",
+  "bos_token": null,
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<|endoftext|>",
+  "errors": "replace",
+  "image_token": "<|image_pad|>",
+  "is_local": false,
+  "model_max_length": 262144,
+  "model_specific_special_tokens": {
+    "audio_bos_token": "<|audio_start|>",
+    "audio_eos_token": "<|audio_end|>",
+    "audio_token": "<|audio_pad|>",
+    "image_token": "<|image_pad|>",
+    "video_token": "<|video_pad|>",
+    "vision_bos_token": "<|vision_start|>",
+    "vision_eos_token": "<|vision_end|>"
+  },
+  "pad_token": "<|endoftext|>",
+  "pretokenize_regex": "(?i:'s|'t|'re|'ve|'m|'ll|'d)|[^\\r\\n\\p{L}\\p{N}]?[\\p{L}\\p{M}]+|\\p{N}| ?[^\\s\\p{L}\\p{M}\\p{N}]+[\\r\\n]*|\\s*[\\r\\n]+|\\s+(?!\\S)|\\s+",
+  "split_special_tokens": false,
+  "tokenizer_class": "TokenizersBackend",
+  "unk_token": null,
+  "video_token": "<|video_pad|>",
+  "vision_bos_token": "<|vision_start|>",
+  "vision_eos_token": "<|vision_end|>"
+}

overgeneralisation_original_Estonian/Qwen3.5-4B-Base_overgeneralisation_splits_original_features_train_overgeneralisation_splits_original_features_test2/checkpoint-1040/trainer_state.json

@@ -0,0 +1,1126 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 2.506940253470127,
+  "eval_steps": 20,
+  "global_step": 1040,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "entropy": 2.172999547421932,
+      "epoch": 0.04828002414001207,
+      "grad_norm": 2.5071208477020264,
+      "learning_rate": 1.7227585487708594e-05,
+      "loss": 1.9347333908081055,
+      "mean_token_accuracy": 0.6086816020309925,
+      "num_tokens": 47778.0,
+      "step": 20
+    },
+    {
+      "epoch": 0.04828002414001207,
+      "eval_entropy": 1.6909754791956269,
+      "eval_loss": 1.4728385210037231,
+      "eval_mean_token_accuracy": 0.6773593797442619,
+      "eval_num_tokens": 47778.0,
+      "eval_runtime": 90.1265,
+      "eval_samples_per_second": 15.756,
+      "eval_steps_per_second": 1.975,
+      "step": 20
+    },
+    {
+      "entropy": 1.2014626950025558,
+      "epoch": 0.09656004828002414,
+      "grad_norm": 1.5074543952941895,
+      "learning_rate": 3.5361886001086065e-05,
+      "loss": 1.0867146492004394,
+      "mean_token_accuracy": 0.7285927519202232,
+      "num_tokens": 95981.0,
+      "step": 40
+    },
+    {
+      "epoch": 0.09656004828002414,
+      "eval_entropy": 1.0304168352250302,
+      "eval_loss": 0.9211422204971313,
+      "eval_mean_token_accuracy": 0.7613122848312507,
+      "eval_num_tokens": 95981.0,
+      "eval_runtime": 88.6338,
+      "eval_samples_per_second": 16.021,
+      "eval_steps_per_second": 2.008,
+      "step": 40
+    },
+    {
+      "entropy": 0.9454198732972146,
+      "epoch": 0.14484007242003621,
+      "grad_norm": 1.7578108310699463,
+      "learning_rate": 5.349618651446354e-05,
+      "loss": 0.8447297096252442,
+      "mean_token_accuracy": 0.7723424732685089,
+      "num_tokens": 142784.0,
+      "step": 60
+    },
+    {
+      "epoch": 0.14484007242003621,
+      "eval_entropy": 0.8647834474451086,
+      "eval_loss": 0.8240737318992615,
+      "eval_mean_token_accuracy": 0.7821227014064789,
+      "eval_num_tokens": 142784.0,
+      "eval_runtime": 88.0135,
+      "eval_samples_per_second": 16.134,
+      "eval_steps_per_second": 2.022,
+      "step": 60
+    },
+    {
+      "entropy": 0.8963550612330436,
+      "epoch": 0.19312009656004828,
+      "grad_norm": 1.1464931964874268,
+      "learning_rate": 7.1630487027841e-05,
+      "loss": 0.7979836463928223,
+      "mean_token_accuracy": 0.7822489373385906,
+      "num_tokens": 187338.0,
+      "step": 80
+    },
+    {
+      "epoch": 0.19312009656004828,
+      "eval_entropy": 0.8663296133614657,
+      "eval_loss": 0.7752988338470459,
+      "eval_mean_token_accuracy": 0.7885957012015782,
+      "eval_num_tokens": 187338.0,
+      "eval_runtime": 88.284,
+      "eval_samples_per_second": 16.084,
+      "eval_steps_per_second": 2.016,
+      "step": 80
+    },
+    {
+      "entropy": 0.8265103206038475,
+      "epoch": 0.24140012070006034,
+      "grad_norm": 1.0106662511825562,
+      "learning_rate": 8.976478754121848e-05,
+      "loss": 0.7297060012817382,
+      "mean_token_accuracy": 0.7955138415098191,
+      "num_tokens": 236038.0,
+      "step": 100
+    },
+    {
+      "epoch": 0.24140012070006034,
+      "eval_entropy": 0.8122169536151244,
+      "eval_loss": 0.754906177520752,
+      "eval_mean_token_accuracy": 0.7917155255092664,
+      "eval_num_tokens": 236038.0,
+      "eval_runtime": 87.5754,
+      "eval_samples_per_second": 16.215,
+      "eval_steps_per_second": 2.033,
+      "step": 100
+    },
+    {
+      "entropy": 0.8098392844200134,
+      "epoch": 0.28968014484007243,
+      "grad_norm": 1.2238041162490845,
+      "learning_rate": 0.00010789908805459595,
+      "loss": 0.7201489448547364,
+      "mean_token_accuracy": 0.7975016921758652,
+      "num_tokens": 282918.0,
+      "step": 120
+    },
+    {
+      "epoch": 0.28968014484007243,
+      "eval_entropy": 0.8418726462326692,
+      "eval_loss": 0.7393178939819336,
+      "eval_mean_token_accuracy": 0.7936771585700217,
+      "eval_num_tokens": 282918.0,
+      "eval_runtime": 87.1597,
+      "eval_samples_per_second": 16.292,
+      "eval_steps_per_second": 2.042,
+      "step": 120
+    },
+    {
+      "entropy": 0.8080591082572937,
+      "epoch": 0.33796016898008446,
+      "grad_norm": 1.011591911315918,
+      "learning_rate": 0.00012603338856797343,
+      "loss": 0.7147697925567627,
+      "mean_token_accuracy": 0.7947175532579422,
+      "num_tokens": 331305.0,
+      "step": 140
+    },
+    {
+      "epoch": 0.33796016898008446,
+      "eval_entropy": 0.7894941431083037,
+      "eval_loss": 0.720219075679779,
+      "eval_mean_token_accuracy": 0.801899236239744,
+      "eval_num_tokens": 331305.0,
+      "eval_runtime": 87.1493,
+      "eval_samples_per_second": 16.294,
+      "eval_steps_per_second": 2.042,
+      "step": 140
+    },
+    {
+      "entropy": 0.7957205109298229,
+      "epoch": 0.38624019312009655,
+      "grad_norm": 0.8481830954551697,
+      "learning_rate": 0.00014416768908135088,
+      "loss": 0.7050027847290039,
+      "mean_token_accuracy": 0.7995368830859662,
+      "num_tokens": 375876.0,
+      "step": 160
+    },
+    {
+      "epoch": 0.38624019312009655,
+      "eval_entropy": 0.7746140591883928,
+      "eval_loss": 0.7175475358963013,
+      "eval_mean_token_accuracy": 0.7994848955213354,
+      "eval_num_tokens": 375876.0,
+      "eval_runtime": 86.655,
+      "eval_samples_per_second": 16.387,
+      "eval_steps_per_second": 2.054,
+      "step": 160
+    },
+    {
+      "entropy": 0.779565304517746,
+      "epoch": 0.43452021726010864,
+      "grad_norm": 1.032624363899231,
+      "learning_rate": 0.00016230198959472835,
+      "loss": 0.6978332042694092,
+      "mean_token_accuracy": 0.8015618294477462,
+      "num_tokens": 423913.0,
+      "step": 180
+    },
+    {
+      "epoch": 0.43452021726010864,
+      "eval_entropy": 0.7455583871080634,
+      "eval_loss": 0.7118850350379944,
+      "eval_mean_token_accuracy": 0.7999682195400923,
+      "eval_num_tokens": 423913.0,
+      "eval_runtime": 86.7842,
+      "eval_samples_per_second": 16.362,
+      "eval_steps_per_second": 2.051,
+      "step": 180
+    },
+    {
+      "entropy": 0.7810172818601131,
+      "epoch": 0.4828002414001207,
+      "grad_norm": 0.9276663064956665,
+      "learning_rate": 0.00018043629010810582,
+      "loss": 0.6937861442565918,
+      "mean_token_accuracy": 0.802893053740263,
+      "num_tokens": 472395.0,
+      "step": 200
+    },
+    {
+      "epoch": 0.4828002414001207,
+      "eval_entropy": 0.7650193290764027,
+      "eval_loss": 0.7020567059516907,
+      "eval_mean_token_accuracy": 0.8045302153973097,
+      "eval_num_tokens": 472395.0,
+      "eval_runtime": 88.326,
+      "eval_samples_per_second": 16.077,
+      "eval_steps_per_second": 2.015,
+      "step": 200
+    },
+    {
+      "entropy": 0.7862060204148292,
+      "epoch": 0.5310802655401328,
+      "grad_norm": 0.8195295929908752,
+      "learning_rate": 0.0001985705906214833,
+      "loss": 0.6878782272338867,
+      "mean_token_accuracy": 0.8027168907225132,
+      "num_tokens": 517049.0,
+      "step": 220
+    },
+    {
+      "epoch": 0.5310802655401328,
+      "eval_entropy": 0.7656565583154057,
+      "eval_loss": 0.6973932981491089,
+      "eval_mean_token_accuracy": 0.8047967278555538,
+      "eval_num_tokens": 517049.0,
+      "eval_runtime": 87.7698,
+      "eval_samples_per_second": 16.179,
+      "eval_steps_per_second": 2.028,
+      "step": 220
+    },
+    {
+      "entropy": 0.7611534461379051,
+      "epoch": 0.5793602896801449,
+      "grad_norm": 0.7499712109565735,
+      "learning_rate": 0.00021670489113486077,
+      "loss": 0.681937837600708,
+      "mean_token_accuracy": 0.8041978515684605,
+      "num_tokens": 562836.0,
+      "step": 240
+    },
+    {
+      "epoch": 0.5793602896801449,
+      "eval_entropy": 0.7599442380197933,
+      "eval_loss": 0.7002046704292297,
+      "eval_mean_token_accuracy": 0.8051861517884759,
+      "eval_num_tokens": 562836.0,
+      "eval_runtime": 87.9239,
+      "eval_samples_per_second": 16.15,
+      "eval_steps_per_second": 2.024,
+      "step": 240
+    },
+    {
+      "entropy": 0.7800783462822437,
+      "epoch": 0.627640313820157,
+      "grad_norm": 0.8645269274711609,
+      "learning_rate": 0.00023483919164823825,
+      "loss": 0.6954500675201416,
+      "mean_token_accuracy": 0.8031484372913837,
+      "num_tokens": 606612.0,
+      "step": 260
+    },
+    {
+      "epoch": 0.627640313820157,
+      "eval_entropy": 0.7498679090751691,
+      "eval_loss": 0.7046149373054504,
+      "eval_mean_token_accuracy": 0.802388215667746,
+      "eval_num_tokens": 606612.0,
+      "eval_runtime": 88.2468,
+      "eval_samples_per_second": 16.091,
+      "eval_steps_per_second": 2.017,
+      "step": 260
+    },
+    {
+      "entropy": 0.7586533665657044,
+      "epoch": 0.6759203379601689,
+      "grad_norm": 0.741596519947052,
+      "learning_rate": 0.0002529734921616157,
+      "loss": 0.690476369857788,
+      "mean_token_accuracy": 0.8046923100948333,
+      "num_tokens": 655920.0,
+      "step": 280
+    },
+    {
+      "epoch": 0.6759203379601689,
+      "eval_entropy": 0.7714274066887544,
+      "eval_loss": 0.6992160677909851,
+      "eval_mean_token_accuracy": 0.8042646482419432,
+      "eval_num_tokens": 655920.0,
+      "eval_runtime": 88.1754,
+      "eval_samples_per_second": 16.104,
+      "eval_steps_per_second": 2.019,
+      "step": 280
+    },
+    {
+      "entropy": 0.7585848286747933,
+      "epoch": 0.724200362100181,
+      "grad_norm": 1.0778124332427979,
+      "learning_rate": 0.00027110779267499317,
+      "loss": 0.6790409564971924,
+      "mean_token_accuracy": 0.8060351841151714,
+      "num_tokens": 702638.0,
+      "step": 300
+    },
+    {
+      "epoch": 0.724200362100181,
+      "eval_entropy": 0.7380402811457601,
+      "eval_loss": 0.7054820656776428,
+      "eval_mean_token_accuracy": 0.8003534738267406,
+      "eval_num_tokens": 702638.0,
+      "eval_runtime": 88.238,
+      "eval_samples_per_second": 16.093,
+      "eval_steps_per_second": 2.017,
+      "step": 300
+    },
+    {
+      "entropy": 0.7696981988847256,
+      "epoch": 0.7724803862401931,
+      "grad_norm": 0.8408658504486084,
+      "learning_rate": 0.00028924209318837064,
+      "loss": 0.7070387363433838,
+      "mean_token_accuracy": 0.8035947173833847,
+      "num_tokens": 749377.0,
+      "step": 320
+    },
+    {
+      "epoch": 0.7724803862401931,
+      "eval_entropy": 0.7897409023193831,
+      "eval_loss": 0.7015668749809265,
+      "eval_mean_token_accuracy": 0.805306687113944,
+      "eval_num_tokens": 749377.0,
+      "eval_runtime": 87.9395,
+      "eval_samples_per_second": 16.147,
+      "eval_steps_per_second": 2.024,
+      "step": 320
+    },
+    {
+      "entropy": 0.762337189912796,
+      "epoch": 0.8207604103802052,
+      "grad_norm": 0.9843310117721558,
+      "learning_rate": 0.0003073763937017481,
+      "loss": 0.7021088123321533,
+      "mean_token_accuracy": 0.8029616877436638,
+      "num_tokens": 798874.0,
+      "step": 340
+    },
+    {
+      "epoch": 0.8207604103802052,
+      "eval_entropy": 0.7210077121016685,
+      "eval_loss": 0.7082746624946594,
+      "eval_mean_token_accuracy": 0.8049283824609906,
+      "eval_num_tokens": 798874.0,
+      "eval_runtime": 88.1706,
+      "eval_samples_per_second": 16.105,
+      "eval_steps_per_second": 2.019,
+      "step": 340
+    },
+    {
+      "entropy": 0.7879025422036647,
+      "epoch": 0.8690404345202173,
+      "grad_norm": 0.9713814854621887,
+      "learning_rate": 0.0003255106942151256,
+      "loss": 0.714624547958374,
+      "mean_token_accuracy": 0.79987031519413,
+      "num_tokens": 840825.0,
+      "step": 360
+    },
+    {
+      "epoch": 0.8690404345202173,
+      "eval_entropy": 0.7866930489459735,
+      "eval_loss": 0.7145519852638245,
+      "eval_mean_token_accuracy": 0.8015657988157165,
+      "eval_num_tokens": 840825.0,
+      "eval_runtime": 88.3746,
+      "eval_samples_per_second": 16.068,
+      "eval_steps_per_second": 2.014,
+      "step": 360
+    },
+    {
+      "entropy": 0.7844726584851742,
+      "epoch": 0.9173204586602294,
+      "grad_norm": 1.2028311491012573,
+      "learning_rate": 0.00034364499472850306,
+      "loss": 0.7040786266326904,
+      "mean_token_accuracy": 0.7981885217130185,
+      "num_tokens": 882885.0,
+      "step": 380
+    },
+    {
+      "epoch": 0.9173204586602294,
+      "eval_entropy": 0.7767668339643585,
+      "eval_loss": 0.7073574066162109,
+      "eval_mean_token_accuracy": 0.8040851833445303,
+      "eval_num_tokens": 882885.0,
+      "eval_runtime": 88.3828,
+      "eval_samples_per_second": 16.066,
+      "eval_steps_per_second": 2.014,
+      "step": 380
+    },
+    {
+      "entropy": 0.7699790760874748,
+      "epoch": 0.9656004828002414,
+      "grad_norm": 0.9397398233413696,
+      "learning_rate": 0.00036177929524188054,
+      "loss": 0.7040313720703125,
+      "mean_token_accuracy": 0.8034727744758129,
+      "num_tokens": 927456.0,
+      "step": 400
+    },
+    {
+      "epoch": 0.9656004828002414,
+      "eval_entropy": 0.7749103545472863,
+      "eval_loss": 0.7093513011932373,
+      "eval_mean_token_accuracy": 0.8019482901926791,
+      "eval_num_tokens": 927456.0,
+      "eval_runtime": 86.2112,
+      "eval_samples_per_second": 16.471,
+      "eval_steps_per_second": 2.065,
+      "step": 400
+    },
+    {
+      "entropy": 0.7672389172888422,
+      "epoch": 1.012070006035003,
+      "grad_norm": 0.99210524559021,
+      "learning_rate": 0.00037628567078152296,
+      "loss": 0.6913118362426758,
+      "mean_token_accuracy": 0.8042210944287189,
+      "num_tokens": 972662.0,
+      "step": 420
+    },
+    {
+      "epoch": 1.012070006035003,
+      "eval_entropy": 0.7294771229283193,
+      "eval_loss": 0.7171520590782166,
+      "eval_mean_token_accuracy": 0.8015128585059991,
+      "eval_num_tokens": 972662.0,
+      "eval_runtime": 88.037,
+      "eval_samples_per_second": 16.13,
+      "eval_steps_per_second": 2.022,
+      "step": 420
+    },
+    {
+      "entropy": 0.7143943183124065,
+      "epoch": 1.060350030175015,
+      "grad_norm": 1.087638258934021,
+      "learning_rate": 0.0003762484015601069,
+      "loss": 0.642704439163208,
+      "mean_token_accuracy": 0.8124966286122799,
+      "num_tokens": 1022525.0,
+      "step": 440
+    },
+    {
+      "epoch": 1.060350030175015,
+      "eval_entropy": 0.743511886744017,
+      "eval_loss": 0.7116662859916687,
+      "eval_mean_token_accuracy": 0.8037210672758939,
+      "eval_num_tokens": 1022525.0,
+      "eval_runtime": 87.9931,
+      "eval_samples_per_second": 16.138,
+      "eval_steps_per_second": 2.023,
+      "step": 440
+    },
+    {
+      "entropy": 0.7237346574664116,
+      "epoch": 1.1086300543150271,
+      "grad_norm": 0.9365553259849548,
+      "learning_rate": 0.0003761579008392123,
+      "loss": 0.652644681930542,
+      "mean_token_accuracy": 0.8107496216893196,
+      "num_tokens": 1069194.0,
+      "step": 460
+    },
+    {
+      "epoch": 1.1086300543150271,
+      "eval_entropy": 0.7372823839776972,
+      "eval_loss": 0.7238303422927856,
+      "eval_mean_token_accuracy": 0.8026230124275336,
+      "eval_num_tokens": 1069194.0,
+      "eval_runtime": 87.6778,
+      "eval_samples_per_second": 16.196,
+      "eval_steps_per_second": 2.03,
+      "step": 460
+    },
+    {
+      "entropy": 0.7202137842774391,
+      "epoch": 1.1569100784550392,
+      "grad_norm": 0.7998443841934204,
+      "learning_rate": 0.0003760141942294726,
+      "loss": 0.6485219478607178,
+      "mean_token_accuracy": 0.8092714451253414,
+      "num_tokens": 1117570.0,
+      "step": 480
+    },
+    {
+      "epoch": 1.1569100784550392,
+      "eval_entropy": 0.71547647942318,
+      "eval_loss": 0.7124277949333191,
+      "eval_mean_token_accuracy": 0.8013174219077892,
+      "eval_num_tokens": 1117570.0,
+      "eval_runtime": 87.4387,
+      "eval_samples_per_second": 16.24,
+      "eval_steps_per_second": 2.036,
+      "step": 480
+    },
+    {
+      "entropy": 0.7270819500088692,
+      "epoch": 1.2051901025950513,
+      "grad_norm": 0.9470787644386292,
+      "learning_rate": 0.00037581732239815854,
+      "loss": 0.66352219581604,
+      "mean_token_accuracy": 0.8102593503892421,
+      "num_tokens": 1161717.0,
+      "step": 500
+    },
+    {
+      "epoch": 1.2051901025950513,
+      "eval_entropy": 0.7791565897759427,
+      "eval_loss": 0.7179726958274841,
+      "eval_mean_token_accuracy": 0.7948898328154275,
+      "eval_num_tokens": 1161717.0,
+      "eval_runtime": 87.9927,
+      "eval_samples_per_second": 16.138,
+      "eval_steps_per_second": 2.023,
+      "step": 500
+    },
+    {
+      "entropy": 0.7320936664938926,
+      "epoch": 1.2534701267350634,
+      "grad_norm": 1.0220379829406738,
+      "learning_rate": 0.0003755673410576695,
+      "loss": 0.6581549644470215,
+      "mean_token_accuracy": 0.8098709337413311,
+      "num_tokens": 1209390.0,
+      "step": 520
+    },
+    {
+      "epoch": 1.2534701267350634,
+      "eval_entropy": 0.7710678225822662,
+      "eval_loss": 0.708797812461853,
+      "eval_mean_token_accuracy": 0.8039572778042783,
+      "eval_num_tokens": 1209390.0,
+      "eval_runtime": 88.1193,
+      "eval_samples_per_second": 16.115,
+      "eval_steps_per_second": 2.02,
+      "step": 520
+    },
+    {
+      "entropy": 0.7281674664467573,
+      "epoch": 1.3017501508750755,
+      "grad_norm": 0.8775188326835632,
+      "learning_rate": 0.000375264320949768,
+      "loss": 0.6631278991699219,
+      "mean_token_accuracy": 0.8101261422038079,
+      "num_tokens": 1256611.0,
+      "step": 540
+    },
+    {
+      "epoch": 1.3017501508750755,
+      "eval_entropy": 0.7199799800187014,
+      "eval_loss": 0.6981531381607056,
+      "eval_mean_token_accuracy": 0.8071265254127845,
+      "eval_num_tokens": 1256611.0,
+      "eval_runtime": 88.165,
+      "eval_samples_per_second": 16.106,
+      "eval_steps_per_second": 2.019,
+      "step": 540
+    },
+    {
+      "entropy": 0.722860773652792,
+      "epoch": 1.3500301750150876,
+      "grad_norm": 0.8949288129806519,
+      "learning_rate": 0.00037490834782556,
+      "loss": 0.6639598846435547,
+      "mean_token_accuracy": 0.8105649061501026,
+      "num_tokens": 1302325.0,
+      "step": 560
+    },
+    {
+      "epoch": 1.3500301750150876,
+      "eval_entropy": 0.7134760032209118,
+      "eval_loss": 0.7029948234558105,
+      "eval_mean_token_accuracy": 0.8062985880991046,
+      "eval_num_tokens": 1302325.0,
+      "eval_runtime": 88.1286,
+      "eval_samples_per_second": 16.113,
+      "eval_steps_per_second": 2.02,
+      "step": 560
+    },
+    {
+      "entropy": 0.7299126699566841,
+      "epoch": 1.3983101991550995,
+      "grad_norm": 1.0323039293289185,
+      "learning_rate": 0.0003744995224212291,
+      "loss": 0.6535763263702392,
+      "mean_token_accuracy": 0.8120224848389626,
+      "num_tokens": 1343844.0,
+      "step": 580
+    },
+    {
+      "epoch": 1.3983101991550995,
+      "eval_entropy": 0.7169701182440426,
+      "eval_loss": 0.7067857980728149,
+      "eval_mean_token_accuracy": 0.8040432936689826,
+      "eval_num_tokens": 1343844.0,
+      "eval_runtime": 88.2252,
+      "eval_samples_per_second": 16.095,
+      "eval_steps_per_second": 2.018,
+      "step": 580
+    },
+    {
+      "entropy": 0.7281742256134749,
+      "epoch": 1.4465902232951118,
+      "grad_norm": 1.3370424509048462,
+      "learning_rate": 0.00037403796042952863,
+      "loss": 0.6637272357940673,
+      "mean_token_accuracy": 0.8118261776864528,
+      "num_tokens": 1389471.0,
+      "step": 600
+    },
+    {
+      "epoch": 1.4465902232951118,
+      "eval_entropy": 0.7378843212395572,
+      "eval_loss": 0.7038390040397644,
+      "eval_mean_token_accuracy": 0.8059229371922739,
+      "eval_num_tokens": 1389471.0,
+      "eval_runtime": 87.9301,
+      "eval_samples_per_second": 16.149,
+      "eval_steps_per_second": 2.024,
+      "step": 600
+    },
+    {
+      "entropy": 0.7388614989817143,
+      "epoch": 1.4948702474351236,
+      "grad_norm": 0.8464171290397644,
+      "learning_rate": 0.00037352379246704257,
+      "loss": 0.6604740142822265,
+      "mean_token_accuracy": 0.8102974124252796,
+      "num_tokens": 1435975.0,
+      "step": 620
+    },
+    {
+      "epoch": 1.4948702474351236,
+      "eval_entropy": 0.7126847046814607,
+      "eval_loss": 0.6950703263282776,
+      "eval_mean_token_accuracy": 0.8085885392815879,
+      "eval_num_tokens": 1435975.0,
+      "eval_runtime": 88.2732,
+      "eval_samples_per_second": 16.086,
+      "eval_steps_per_second": 2.016,
+      "step": 620
+    },
+    {
+      "entropy": 0.7258916139602661,
+      "epoch": 1.5431502715751357,
+      "grad_norm": 1.1780085563659668,
+      "learning_rate": 0.0003729571640372223,
+      "loss": 0.6529891014099121,
+      "mean_token_accuracy": 0.8085981778800487,
+      "num_tokens": 1482035.0,
+      "step": 640
+    },
+    {
+      "epoch": 1.5431502715751357,
+      "eval_entropy": 0.7096509638797032,
+      "eval_loss": 0.690051257610321,
+      "eval_mean_token_accuracy": 0.8100520241796301,
+      "eval_num_tokens": 1482035.0,
+      "eval_runtime": 88.3153,
+      "eval_samples_per_second": 16.079,
+      "eval_steps_per_second": 2.016,
+      "step": 640
+    },
+    {
+      "entropy": 0.7151173129677773,
+      "epoch": 1.5914302957151478,
+      "grad_norm": 1.0357667207717896,
+      "learning_rate": 0.0003723382354892108,
+      "loss": 0.6451474189758301,
+      "mean_token_accuracy": 0.8136902332305909,
+      "num_tokens": 1528119.0,
+      "step": 660
+    },
+    {
+      "epoch": 1.5914302957151478,
+      "eval_entropy": 0.7060252652409371,
+      "eval_loss": 0.6899483799934387,
+      "eval_mean_token_accuracy": 0.810834194502134,
+      "eval_num_tokens": 1528119.0,
+      "eval_runtime": 87.8848,
+      "eval_samples_per_second": 16.158,
+      "eval_steps_per_second": 2.025,
+      "step": 660
+    },
+    {
+      "entropy": 0.7141377851366997,
+      "epoch": 1.63971031985516,
+      "grad_norm": 0.8391667008399963,
+      "learning_rate": 0.000371667181972466,
+      "loss": 0.6481579780578614,
+      "mean_token_accuracy": 0.8159015104174614,
+      "num_tokens": 1573661.0,
+      "step": 680
+    },
+    {
+      "epoch": 1.63971031985516,
+      "eval_entropy": 0.7598993319473909,
+      "eval_loss": 0.681204080581665,
+      "eval_mean_token_accuracy": 0.8100194586126992,
+      "eval_num_tokens": 1573661.0,
+      "eval_runtime": 88.2415,
+      "eval_samples_per_second": 16.092,
+      "eval_steps_per_second": 2.017,
+      "step": 680
+    },
+    {
+      "entropy": 0.7235980287194252,
+      "epoch": 1.687990343995172,
+      "grad_norm": 0.8197196125984192,
+      "learning_rate": 0.00037094419338719537,
+      "loss": 0.6582849025726318,
+      "mean_token_accuracy": 0.8114834539592266,
+      "num_tokens": 1621714.0,
+      "step": 700
+    },
+    {
+      "epoch": 1.687990343995172,
+      "eval_entropy": 0.7166854568411795,
+      "eval_loss": 0.679177463054657,
+      "eval_mean_token_accuracy": 0.8119039877077167,
+      "eval_num_tokens": 1621714.0,
+      "eval_runtime": 88.0957,
+      "eval_samples_per_second": 16.119,
+      "eval_steps_per_second": 2.021,
+      "step": 700
+    },
+    {
+      "entropy": 0.7278190158307553,
+      "epoch": 1.736270368135184,
+      "grad_norm": 0.7820602655410767,
+      "learning_rate": 0.0003701694743306164,
+      "loss": 0.6633153438568116,
+      "mean_token_accuracy": 0.8123249113559723,
+      "num_tokens": 1672118.0,
+      "step": 720
+    },
+    {
+      "epoch": 1.736270368135184,
+      "eval_entropy": 0.6980850314826108,
+      "eval_loss": 0.6775653958320618,
+      "eval_mean_token_accuracy": 0.8116872906684875,
+      "eval_num_tokens": 1672118.0,
+      "eval_runtime": 87.9795,
+      "eval_samples_per_second": 16.14,
+      "eval_steps_per_second": 2.023,
+      "step": 720
+    },
+    {
+      "entropy": 0.7149631556123495,
+      "epoch": 1.7845503922751962,
+      "grad_norm": 1.0139760971069336,
+      "learning_rate": 0.00036934324403905795,
+      "loss": 0.6468976020812989,
+      "mean_token_accuracy": 0.8117561548948288,
+      "num_tokens": 1720465.0,
+      "step": 740
+    },
+    {
+      "epoch": 1.7845503922751962,
+      "eval_entropy": 0.69009849094273,
+      "eval_loss": 0.6731093525886536,
+      "eval_mean_token_accuracy": 0.81281964196248,
+      "eval_num_tokens": 1720465.0,
+      "eval_runtime": 87.7412,
+      "eval_samples_per_second": 16.184,
+      "eval_steps_per_second": 2.029,
+      "step": 740
+    },
+    {
+      "entropy": 0.7226049453020096,
+      "epoch": 1.832830416415208,
+      "grad_norm": 1.5040546655654907,
+      "learning_rate": 0.0003684657363259193,
+      "loss": 0.6385328769683838,
+      "mean_token_accuracy": 0.8132639616727829,
+      "num_tokens": 1764440.0,
+      "step": 760
+    },
+    {
+      "epoch": 1.832830416415208,
+      "eval_entropy": 0.7125169618076153,
+      "eval_loss": 0.6864680647850037,
+      "eval_mean_token_accuracy": 0.8049418089095126,
+      "eval_num_tokens": 1764440.0,
+      "eval_runtime": 87.9429,
+      "eval_samples_per_second": 16.147,
+      "eval_steps_per_second": 2.024,
+      "step": 760
+    },
+    {
+      "entropy": 0.707821810990572,
+      "epoch": 1.8811104405552204,
+      "grad_norm": 0.8604223132133484,
+      "learning_rate": 0.00036753719951550327,
+      "loss": 0.6397049427032471,
+      "mean_token_accuracy": 0.8137877315282822,
+      "num_tokens": 1810828.0,
+      "step": 780
+    },
+    {
+      "epoch": 1.8811104405552204,
+      "eval_entropy": 0.6492102551326323,
+      "eval_loss": 0.6743726134300232,
+      "eval_mean_token_accuracy": 0.8130356572317273,
+      "eval_num_tokens": 1810828.0,
+      "eval_runtime": 87.8684,
+      "eval_samples_per_second": 16.161,
+      "eval_steps_per_second": 2.026,
+      "step": 780
+    },
+    {
+      "entropy": 0.6800355084240437,
+      "epoch": 1.9293904646952322,
+      "grad_norm": 0.7270982265472412,
+      "learning_rate": 0.00036655789637274377,
+      "loss": 0.6267755031585693,
+      "mean_token_accuracy": 0.8180945307016373,
+      "num_tokens": 1857600.0,
+      "step": 800
+    },
+    {
+      "epoch": 1.9293904646952322,
+      "eval_entropy": 0.6727538651294922,
+      "eval_loss": 0.6696028709411621,
+      "eval_mean_token_accuracy": 0.8116443090224534,
+      "eval_num_tokens": 1857600.0,
+      "eval_runtime": 88.2449,
+      "eval_samples_per_second": 16.092,
+      "eval_steps_per_second": 2.017,
+      "step": 800
+    },
+    {
+      "entropy": 0.6746691606938839,
+      "epoch": 1.9776704888352445,
+      "grad_norm": 0.9771540760993958,
+      "learning_rate": 0.0003655281040288461,
+      "loss": 0.6326992988586426,
+      "mean_token_accuracy": 0.8179976396262646,
+      "num_tokens": 1901998.0,
+      "step": 820
+    },
+    {
+      "epoch": 1.9776704888352445,
+      "eval_entropy": 0.6636319716324967,
+      "eval_loss": 0.6714363098144531,
+      "eval_mean_token_accuracy": 0.8126635943235976,
+      "eval_num_tokens": 1901998.0,
+      "eval_runtime": 87.7666,
+      "eval_samples_per_second": 16.179,
+      "eval_steps_per_second": 2.028,
+      "step": 820
+    },
+    {
+      "entropy": 0.6319125778489298,
+      "epoch": 2.024140012070006,
+      "grad_norm": 0.8380444645881653,
+      "learning_rate": 0.0003644481139028622,
+      "loss": 0.5780903816223144,
+      "mean_token_accuracy": 0.8250188401767186,
+      "num_tokens": 1947553.0,
+      "step": 840
+    },
+    {
+      "epoch": 2.024140012070006,
+      "eval_entropy": 0.6627151989535,
+      "eval_loss": 0.6774935722351074,
+      "eval_mean_token_accuracy": 0.8140510819601209,
+      "eval_num_tokens": 1947553.0,
+      "eval_runtime": 87.8287,
+      "eval_samples_per_second": 16.168,
+      "eval_steps_per_second": 2.027,
+      "step": 840
+    },
+    {
+      "entropy": 0.6059147048741579,
+      "epoch": 2.0724200362100182,
+      "grad_norm": 0.9745152592658997,
+      "learning_rate": 0.0003633182316192229,
+      "loss": 0.546702527999878,
+      "mean_token_accuracy": 0.8358186542987823,
+      "num_tokens": 1995234.0,
+      "step": 860
+    },
+    {
+      "epoch": 2.0724200362100182,
+      "eval_entropy": 0.6620710384979677,
+      "eval_loss": 0.6769291162490845,
+      "eval_mean_token_accuracy": 0.8135082835561773,
+      "eval_num_tokens": 1995234.0,
+      "eval_runtime": 88.0008,
+      "eval_samples_per_second": 16.136,
+      "eval_steps_per_second": 2.023,
+      "step": 860
+    },
+    {
+      "entropy": 0.6076117843389511,
+      "epoch": 2.12070006035003,
+      "grad_norm": 0.6914380192756653,
+      "learning_rate": 0.0003621387769212491,
+      "loss": 0.5324535846710206,
+      "mean_token_accuracy": 0.8366568259894848,
+      "num_tokens": 2040639.0,
+      "step": 880
+    },
+    {
+      "epoch": 2.12070006035003,
+      "eval_entropy": 0.6592224213514435,
+      "eval_loss": 0.670383632183075,
+      "eval_mean_token_accuracy": 0.8162150088320957,
+      "eval_num_tokens": 2040639.0,
+      "eval_runtime": 88.324,
+      "eval_samples_per_second": 16.077,
+      "eval_steps_per_second": 2.015,
+      "step": 880
+    },
+    {
+      "entropy": 0.6158072650432587,
+      "epoch": 2.1689800844900424,
+      "grad_norm": 0.9649198055267334,
+      "learning_rate": 0.0003609100835806688,
+      "loss": 0.5564568996429443,
+      "mean_token_accuracy": 0.8329654954373836,
+      "num_tokens": 2090604.0,
+      "step": 900
+    },
+    {
+      "epoch": 2.1689800844900424,
+      "eval_entropy": 0.608588819758276,
+      "eval_loss": 0.6745102405548096,
+      "eval_mean_token_accuracy": 0.8155080542135774,
+      "eval_num_tokens": 2090604.0,
+      "eval_runtime": 87.52,
+      "eval_samples_per_second": 16.225,
+      "eval_steps_per_second": 2.034,
+      "step": 900
+    },
+    {
+      "entropy": 0.5938734326511621,
+      "epoch": 2.2172601086300543,
+      "grad_norm": 0.6915557384490967,
+      "learning_rate": 0.0003596324993031632,
+      "loss": 0.5351875782012939,
+      "mean_token_accuracy": 0.8385631121695042,
+      "num_tokens": 2142246.0,
+      "step": 920
+    },
+    {
+      "epoch": 2.2172601086300543,
+      "eval_entropy": 0.6251631205001574,
+      "eval_loss": 0.6664518117904663,
+      "eval_mean_token_accuracy": 0.81711940417129,
+      "eval_num_tokens": 2142246.0,
+      "eval_runtime": 87.1792,
+      "eval_samples_per_second": 16.288,
+      "eval_steps_per_second": 2.042,
+      "step": 920
+    },
+    {
+      "entropy": 0.623173613846302,
+      "epoch": 2.2655401327700666,
+      "grad_norm": 0.8366610407829285,
+      "learning_rate": 0.00035830638562997063,
+      "loss": 0.5488213539123535,
+      "mean_token_accuracy": 0.8321483485400677,
+      "num_tokens": 2183144.0,
+      "step": 940
+    },
+    {
+      "epoch": 2.2655401327700666,
+      "eval_entropy": 0.6237085649136747,
+      "eval_loss": 0.68075031042099,
+      "eval_mean_token_accuracy": 0.814671738429016,
+      "eval_num_tokens": 2183144.0,
+      "eval_runtime": 88.0652,
+      "eval_samples_per_second": 16.124,
+      "eval_steps_per_second": 2.021,
+      "step": 940
+    },
+    {
+      "entropy": 0.6236911740154027,
+      "epoch": 2.3138201569100785,
+      "grad_norm": 0.7812452912330627,
+      "learning_rate": 0.00035693211783557416,
+      "loss": 0.5696200847625732,
+      "mean_token_accuracy": 0.82991396561265,
+      "num_tokens": 2228098.0,
+      "step": 960
+    },
+    {
+      "epoch": 2.3138201569100785,
+      "eval_entropy": 0.6327033426319615,
+      "eval_loss": 0.6742984652519226,
+      "eval_mean_token_accuracy": 0.8148086091105857,
+      "eval_num_tokens": 2228098.0,
+      "eval_runtime": 88.1711,
+      "eval_samples_per_second": 16.105,
+      "eval_steps_per_second": 2.019,
+      "step": 960
+    },
+    {
+      "entropy": 0.6218426413834095,
+      "epoch": 2.3621001810500903,
+      "grad_norm": 0.8719790577888489,
+      "learning_rate": 0.0003555100848215035,
+      "loss": 0.5582651615142822,
+      "mean_token_accuracy": 0.8323961742222309,
+      "num_tokens": 2273062.0,
+      "step": 980
+    },
+    {
+      "epoch": 2.3621001810500903,
+      "eval_entropy": 0.6338012375858393,
+      "eval_loss": 0.6685822606086731,
+      "eval_mean_token_accuracy": 0.8152180300669724,
+      "eval_num_tokens": 2273062.0,
+      "eval_runtime": 88.101,
+      "eval_samples_per_second": 16.118,
+      "eval_steps_per_second": 2.02,
+      "step": 980
+    },
+    {
+      "entropy": 0.6050681818276644,
+      "epoch": 2.4103802051901027,
+      "grad_norm": 0.9151566028594971,
+      "learning_rate": 0.00035404068900628076,
+      "loss": 0.559452486038208,
+      "mean_token_accuracy": 0.8310242861509323,
+      "num_tokens": 2319305.0,
+      "step": 1000
+    },
+    {
+      "epoch": 2.4103802051901027,
+      "eval_entropy": 0.6634890969549672,
+      "eval_loss": 0.6692465543746948,
+      "eval_mean_token_accuracy": 0.8134353673190213,
+      "eval_num_tokens": 2319305.0,
+      "eval_runtime": 87.7711,
+      "eval_samples_per_second": 16.178,
+      "eval_steps_per_second": 2.028,
+      "step": 1000
+    },
+    {
+      "entropy": 0.6169968567788601,
+      "epoch": 2.4586602293301145,
+      "grad_norm": 0.7081006765365601,
+      "learning_rate": 0.0003525243462115402,
+      "loss": 0.5589694023132324,
+      "mean_token_accuracy": 0.8357502184808254,
+      "num_tokens": 2368592.0,
+      "step": 1020
+    },
+    {
+      "epoch": 2.4586602293301145,
+      "eval_entropy": 0.6126393578695447,
+      "eval_loss": 0.6706712245941162,
+      "eval_mean_token_accuracy": 0.8155577500884452,
+      "eval_num_tokens": 2368592.0,
+      "eval_runtime": 88.1407,
+      "eval_samples_per_second": 16.111,
+      "eval_steps_per_second": 2.019,
+      "step": 1020
+    },
+    {
+      "entropy": 0.6072248566895724,
+      "epoch": 2.506940253470127,
+      "grad_norm": 0.6999370455741882,
+      "learning_rate": 0.0003509614855443561,
+      "loss": 0.5500593185424805,
+      "mean_token_accuracy": 0.8337264291942119,
+      "num_tokens": 2412859.0,
+      "step": 1040
+    },
+    {
+      "epoch": 2.506940253470127,
+      "eval_entropy": 0.6417417074187418,
+      "eval_loss": 0.6651163697242737,
+      "eval_mean_token_accuracy": 0.8159113908081912,
+      "eval_num_tokens": 2412859.0,
+      "eval_runtime": 87.4854,
+      "eval_samples_per_second": 16.231,
+      "eval_steps_per_second": 2.035,
+      "step": 1040
+    }
+  ],
+  "logging_steps": 20,
+  "max_steps": 4150,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 10,
+  "save_steps": 20,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 1.0198081900655616e+17,
+  "train_batch_size": 4,
+  "trial_name": null,
+  "trial_params": null
+}

overgeneralisation_original_Estonian/Qwen3.5-4B-Base_overgeneralisation_splits_original_features_train_overgeneralisation_splits_original_features_test2/checkpoint-1060/README.md

@@ -0,0 +1,209 @@
+---
+base_model: Qwen/Qwen3.5-4B-Base
+library_name: peft
+pipeline_tag: text-generation
+tags:
+- base_model:adapter:Qwen/Qwen3.5-4B-Base
+- lora
+- sft
+- transformers
+- trl
+---
+
+# Model Card for Model ID
+
+<!-- Provide a quick summary of what the model is/does. -->
+
+
+
+## Model Details
+
+### Model Description
+
+<!-- Provide a longer summary of what this model is. -->
+
+
+
+- **Developed by:** [More Information Needed]
+- **Funded by [optional]:** [More Information Needed]
+- **Shared by [optional]:** [More Information Needed]
+- **Model type:** [More Information Needed]
+- **Language(s) (NLP):** [More Information Needed]
+- **License:** [More Information Needed]
+- **Finetuned from model [optional]:** [More Information Needed]
+
+### Model Sources [optional]
+
+<!-- Provide the basic links for the model. -->
+
+- **Repository:** [More Information Needed]
+- **Paper [optional]:** [More Information Needed]
+- **Demo [optional]:** [More Information Needed]
+
+## Uses
+
+<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
+
+### Direct Use
+
+<!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
+
+[More Information Needed]
+
+### Downstream Use [optional]
+
+<!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->
+
+[More Information Needed]
+
+### Out-of-Scope Use
+
+<!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
+
+[More Information Needed]
+
+## Bias, Risks, and Limitations
+
+<!-- This section is meant to convey both technical and sociotechnical limitations. -->
+
+[More Information Needed]
+
+### Recommendations
+
+<!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->
+
+Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
+
+## How to Get Started with the Model
+
+Use the code below to get started with the model.
+
+[More Information Needed]
+
+## Training Details
+
+### Training Data
+
+<!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
+
+[More Information Needed]
+
+### Training Procedure
+
+<!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
+
+#### Preprocessing [optional]
+
+[More Information Needed]
+
+
+#### Training Hyperparameters
+
+- **Training regime:** [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->
+
+#### Speeds, Sizes, Times [optional]
+
+<!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
+
+[More Information Needed]
+
+## Evaluation
+
+<!-- This section describes the evaluation protocols and provides the results. -->
+
+### Testing Data, Factors & Metrics
+
+#### Testing Data
+
+<!-- This should link to a Dataset Card if possible. -->
+
+[More Information Needed]
+
+#### Factors
+
+<!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->
+
+[More Information Needed]
+
+#### Metrics
+
+<!-- These are the evaluation metrics being used, ideally with a description of why. -->
+
+[More Information Needed]
+
+### Results
+
+[More Information Needed]
+
+#### Summary
+
+
+
+## Model Examination [optional]
+
+<!-- Relevant interpretability work for the model goes here -->
+
+[More Information Needed]
+
+## Environmental Impact
+
+<!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly -->
+
+Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700).
+
+- **Hardware Type:** [More Information Needed]
+- **Hours used:** [More Information Needed]
+- **Cloud Provider:** [More Information Needed]
+- **Compute Region:** [More Information Needed]
+- **Carbon Emitted:** [More Information Needed]
+
+## Technical Specifications [optional]
+
+### Model Architecture and Objective
+
+[More Information Needed]
+
+### Compute Infrastructure
+
+[More Information Needed]
+
+#### Hardware
+
+[More Information Needed]
+
+#### Software
+
+[More Information Needed]
+
+## Citation [optional]
+
+<!-- If there is a paper or blog post introducing the model, the APA and Bibtex information for that should go in this section. -->
+
+**BibTeX:**
+
+[More Information Needed]
+
+**APA:**
+
+[More Information Needed]
+
+## Glossary [optional]
+
+<!-- If relevant, include terms and calculations in this section that can help readers understand the model or model card. -->
+
+[More Information Needed]
+
+## More Information [optional]
+
+[More Information Needed]
+
+## Model Card Authors [optional]
+
+[More Information Needed]
+
+## Model Card Contact
+
+[More Information Needed]
+### Framework versions
+
+- PEFT 0.18.1

overgeneralisation_original_Estonian/Qwen3.5-4B-Base_overgeneralisation_splits_original_features_train_overgeneralisation_splits_original_features_test2/checkpoint-1060/adapter_config.json

@@ -0,0 +1,46 @@
+{
+  "alora_invocation_tokens": null,
+  "alpha_pattern": {},
+  "arrow_config": null,
+  "auto_mapping": null,
+  "base_model_name_or_path": "Qwen/Qwen3.5-4B-Base",
+  "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": 128,
+  "lora_bias": false,
+  "lora_dropout": 0.03828026524568501,
+  "megatron_config": null,
+  "megatron_core": "megatron.core",
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "peft_version": "0.18.1",
+  "qalora_group_size": 16,
+  "r": 64,
+  "rank_pattern": {},
+  "revision": null,
+  "target_modules": [
+    "gate_proj",
+    "v_proj",
+    "up_proj",
+    "down_proj",
+    "o_proj",
+    "q_proj",
+    "k_proj"
+  ],
+  "target_parameters": null,
+  "task_type": "CAUSAL_LM",
+  "trainable_token_indices": null,
+  "use_dora": false,
+  "use_qalora": false,
+  "use_rslora": false
+}

overgeneralisation_original_Estonian/Qwen3.5-4B-Base_overgeneralisation_splits_original_features_train_overgeneralisation_splits_original_features_test2/checkpoint-1060/chat_template.jinja

@@ -0,0 +1,154 @@
+{%- set image_count = namespace(value=0) %}
+{%- set video_count = namespace(value=0) %}
+{%- macro render_content(content, do_vision_count, is_system_content=false) %}
+    {%- if content is string %}
+        {{- content }}
+    {%- elif content is iterable and content is not mapping %}
+        {%- for item in content %}
+            {%- if 'image' in item or 'image_url' in item or item.type == 'image' %}
+                {%- if is_system_content %}
+                    {{- raise_exception('System message cannot contain images.') }}
+                {%- endif %}
+                {%- if do_vision_count %}
+                    {%- set image_count.value = image_count.value + 1 %}
+                {%- endif %}
+                {%- if add_vision_id %}
+                    {{- 'Picture ' ~ image_count.value ~ ': ' }}
+                {%- endif %}
+                {{- '<|vision_start|><|image_pad|><|vision_end|>' }}
+            {%- elif 'video' in item or item.type == 'video' %}
+                {%- if is_system_content %}
+                    {{- raise_exception('System message cannot contain videos.') }}
+                {%- endif %}
+                {%- if do_vision_count %}
+                    {%- set video_count.value = video_count.value + 1 %}
+                {%- endif %}
+                {%- if add_vision_id %}
+                    {{- 'Video ' ~ video_count.value ~ ': ' }}
+                {%- endif %}
+                {{- '<|vision_start|><|video_pad|><|vision_end|>' }}
+            {%- elif 'text' in item %}
+                {{- item.text }}
+            {%- else %}
+                {{- raise_exception('Unexpected item type in content.') }}
+            {%- endif %}
+        {%- endfor %}
+    {%- elif content is none or content is undefined %}
+        {{- '' }}
+    {%- else %}
+        {{- raise_exception('Unexpected content type.') }}
+    {%- endif %}
+{%- endmacro %}
+{%- if not messages %}
+    {{- raise_exception('No messages provided.') }}
+{%- endif %}
+{%- if tools and tools is iterable and tools is not mapping %}
+    {{- '<|im_start|>system\n' }}
+    {{- "# Tools\n\nYou have access to the following functions:\n\n<tools>" }}
+    {%- for tool in tools %}
+        {{- "\n" }}
+        {{- tool | tojson }}
+    {%- endfor %}
+    {{- "\n</tools>" }}
+    {{- '\n\nIf you choose to call a function ONLY reply in the following format with NO suffix:\n\n<tool_call>\n<function=example_function_name>\n<parameter=example_parameter_1>\nvalue_1\n</parameter>\n<parameter=example_parameter_2>\nThis is the value for the second parameter\nthat can span\nmultiple lines\n</parameter>\n</function>\n</tool_call>\n\n<IMPORTANT>\nReminder:\n- Function calls MUST follow the specified format: an inner <function=...></function> block must be nested within <tool_call></tool_call> XML tags\n- Required parameters MUST be specified\n- You may provide optional reasoning for your function call in natural language BEFORE the function call, but NOT after\n- If there is no function call available, answer the question like normal with your current knowledge and do not tell the user about function calls\n</IMPORTANT>' }}
+    {%- if messages[0].role == 'system' %}
+        {%- set content = render_content(messages[0].content, false, true)|trim %}
+        {%- if content %}
+            {{- '\n\n' + content }}
+        {%- endif %}
+    {%- endif %}
+    {{- '<|im_end|>\n' }}
+{%- else %}
+    {%- if messages[0].role == 'system' %}
+        {%- set content = render_content(messages[0].content, false, true)|trim %}
+        {{- '<|im_start|>system\n' + content + '<|im_end|>\n' }}
+    {%- endif %}
+{%- endif %}
+{%- set ns = namespace(multi_step_tool=true, last_query_index=messages|length - 1) %}
+{%- for message in messages[::-1] %}
+    {%- set index = (messages|length - 1) - loop.index0 %}
+    {%- if ns.multi_step_tool and message.role == "user" %}
+        {%- set content = render_content(message.content, false)|trim %}
+        {%- if not(content.startswith('<tool_response>') and content.endswith('</tool_response>')) %}
+            {%- set ns.multi_step_tool = false %}
+            {%- set ns.last_query_index = index %}
+        {%- endif %}
+    {%- endif %}
+{%- endfor %}
+{%- if ns.multi_step_tool %}
+    {{- raise_exception('No user query found in messages.') }}
+{%- endif %}
+{%- for message in messages %}
+    {%- set content = render_content(message.content, true)|trim %}
+    {%- if message.role == "system" %}
+        {%- if not loop.first %}
+            {{- raise_exception('System message must be at the beginning.') }}
+        {%- endif %}
+    {%- elif message.role == "user" %}
+        {{- '<|im_start|>' + message.role + '\n' + content + '<|im_end|>' + '\n' }}
+    {%- elif message.role == "assistant" %}
+        {%- set reasoning_content = '' %}
+        {%- if message.reasoning_content is string %}
+            {%- set reasoning_content = message.reasoning_content %}
+        {%- else %}
+            {%- if '</think>' in content %}
+                {%- set reasoning_content = content.split('</think>')[0].rstrip('\n').split('<think>')[-1].lstrip('\n') %}
+                {%- set content = content.split('</think>')[-1].lstrip('\n') %}
+            {%- endif %}
+        {%- endif %}
+        {%- set reasoning_content = reasoning_content|trim %}
+        {%- if loop.index0 > ns.last_query_index %}
+            {{- '<|im_start|>' + message.role + '\n<think>\n' + reasoning_content + '\n</think>\n\n' + content }}
+        {%- else %}
+            {{- '<|im_start|>' + message.role + '\n' + content }}
+        {%- endif %}
+        {%- if message.tool_calls and message.tool_calls is iterable and message.tool_calls is not mapping %}
+            {%- for tool_call in message.tool_calls %}
+                {%- if tool_call.function is defined %}
+                    {%- set tool_call = tool_call.function %}
+                {%- endif %}
+                {%- if loop.first %}
+                    {%- if content|trim %}
+                        {{- '\n\n<tool_call>\n<function=' + tool_call.name + '>\n' }}
+                    {%- else %}
+                        {{- '<tool_call>\n<function=' + tool_call.name + '>\n' }}
+                    {%- endif %}
+                {%- else %}
+                    {{- '\n<tool_call>\n<function=' + tool_call.name + '>\n' }}
+                {%- endif %}
+                {%- if tool_call.arguments is defined %}
+                    {%- for args_name, args_value in tool_call.arguments|items %}
+                        {{- '<parameter=' + args_name + '>\n' }}
+                        {%- set args_value = args_value | tojson | safe if args_value is mapping or (args_value is sequence and args_value is not string) else args_value | string %}
+                        {{- args_value }}
+                        {{- '\n</parameter>\n' }}
+                    {%- endfor %}
+                {%- endif %}
+                {{- '</function>\n</tool_call>' }}
+            {%- endfor %}
+        {%- endif %}
+        {{- '<|im_end|>\n' }}
+    {%- elif message.role == "tool" %}
+        {%- if loop.previtem and loop.previtem.role != "tool" %}
+            {{- '<|im_start|>user' }}
+        {%- endif %}
+        {{- '\n<tool_response>\n' }}
+        {{- content }}
+        {{- '\n</tool_response>' }}
+        {%- if not loop.last and loop.nextitem.role != "tool" %}
+            {{- '<|im_end|>\n' }}
+        {%- elif loop.last %}
+            {{- '<|im_end|>\n' }}
+        {%- endif %}
+    {%- else %}
+        {{- raise_exception('Unexpected message role.') }}
+    {%- endif %}
+{%- endfor %}
+{%- if add_generation_prompt %}
+    {{- '<|im_start|>assistant\n' }}
+    {%- if enable_thinking is defined and enable_thinking is false %}
+        {{- '<think>\n\n</think>\n\n' }}
+    {%- else %}
+        {{- '<think>\n' }}
+    {%- endif %}
+{%- endif %}

overgeneralisation_original_Estonian/Qwen3.5-4B-Base_overgeneralisation_splits_original_features_train_overgeneralisation_splits_original_features_test2/checkpoint-1060/tokenizer_config.json

@@ -0,0 +1,31 @@
+{
+  "add_prefix_space": false,
+  "audio_bos_token": "<|audio_start|>",
+  "audio_eos_token": "<|audio_end|>",
+  "audio_token": "<|audio_pad|>",
+  "backend": "tokenizers",
+  "bos_token": null,
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<|endoftext|>",
+  "errors": "replace",
+  "image_token": "<|image_pad|>",
+  "is_local": false,
+  "model_max_length": 262144,
+  "model_specific_special_tokens": {
+    "audio_bos_token": "<|audio_start|>",
+    "audio_eos_token": "<|audio_end|>",
+    "audio_token": "<|audio_pad|>",
+    "image_token": "<|image_pad|>",
+    "video_token": "<|video_pad|>",
+    "vision_bos_token": "<|vision_start|>",
+    "vision_eos_token": "<|vision_end|>"
+  },
+  "pad_token": "<|endoftext|>",
+  "pretokenize_regex": "(?i:'s|'t|'re|'ve|'m|'ll|'d)|[^\\r\\n\\p{L}\\p{N}]?[\\p{L}\\p{M}]+|\\p{N}| ?[^\\s\\p{L}\\p{M}\\p{N}]+[\\r\\n]*|\\s*[\\r\\n]+|\\s+(?!\\S)|\\s+",
+  "split_special_tokens": false,
+  "tokenizer_class": "TokenizersBackend",
+  "unk_token": null,
+  "video_token": "<|video_pad|>",
+  "vision_bos_token": "<|vision_start|>",
+  "vision_eos_token": "<|vision_end|>"
+}

overgeneralisation_original_Estonian/Qwen3.5-4B-Base_overgeneralisation_splits_original_features_train_overgeneralisation_splits_original_features_test2/checkpoint-1060/trainer_state.json

@@ -0,0 +1,1147 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 2.5552202776101387,
+  "eval_steps": 20,
+  "global_step": 1060,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "entropy": 2.172999547421932,
+      "epoch": 0.04828002414001207,
+      "grad_norm": 2.5071208477020264,
+      "learning_rate": 1.7227585487708594e-05,
+      "loss": 1.9347333908081055,
+      "mean_token_accuracy": 0.6086816020309925,
+      "num_tokens": 47778.0,
+      "step": 20
+    },
+    {
+      "epoch": 0.04828002414001207,
+      "eval_entropy": 1.6909754791956269,
+      "eval_loss": 1.4728385210037231,
+      "eval_mean_token_accuracy": 0.6773593797442619,
+      "eval_num_tokens": 47778.0,
+      "eval_runtime": 90.1265,
+      "eval_samples_per_second": 15.756,
+      "eval_steps_per_second": 1.975,
+      "step": 20
+    },
+    {
+      "entropy": 1.2014626950025558,
+      "epoch": 0.09656004828002414,
+      "grad_norm": 1.5074543952941895,
+      "learning_rate": 3.5361886001086065e-05,
+      "loss": 1.0867146492004394,
+      "mean_token_accuracy": 0.7285927519202232,
+      "num_tokens": 95981.0,
+      "step": 40
+    },
+    {
+      "epoch": 0.09656004828002414,
+      "eval_entropy": 1.0304168352250302,
+      "eval_loss": 0.9211422204971313,
+      "eval_mean_token_accuracy": 0.7613122848312507,
+      "eval_num_tokens": 95981.0,
+      "eval_runtime": 88.6338,
+      "eval_samples_per_second": 16.021,
+      "eval_steps_per_second": 2.008,
+      "step": 40
+    },
+    {
+      "entropy": 0.9454198732972146,
+      "epoch": 0.14484007242003621,
+      "grad_norm": 1.7578108310699463,
+      "learning_rate": 5.349618651446354e-05,
+      "loss": 0.8447297096252442,
+      "mean_token_accuracy": 0.7723424732685089,
+      "num_tokens": 142784.0,
+      "step": 60
+    },
+    {
+      "epoch": 0.14484007242003621,
+      "eval_entropy": 0.8647834474451086,
+      "eval_loss": 0.8240737318992615,
+      "eval_mean_token_accuracy": 0.7821227014064789,
+      "eval_num_tokens": 142784.0,
+      "eval_runtime": 88.0135,
+      "eval_samples_per_second": 16.134,
+      "eval_steps_per_second": 2.022,
+      "step": 60
+    },
+    {
+      "entropy": 0.8963550612330436,
+      "epoch": 0.19312009656004828,
+      "grad_norm": 1.1464931964874268,
+      "learning_rate": 7.1630487027841e-05,
+      "loss": 0.7979836463928223,
+      "mean_token_accuracy": 0.7822489373385906,
+      "num_tokens": 187338.0,
+      "step": 80
+    },
+    {
+      "epoch": 0.19312009656004828,
+      "eval_entropy": 0.8663296133614657,
+      "eval_loss": 0.7752988338470459,
+      "eval_mean_token_accuracy": 0.7885957012015782,
+      "eval_num_tokens": 187338.0,
+      "eval_runtime": 88.284,
+      "eval_samples_per_second": 16.084,
+      "eval_steps_per_second": 2.016,
+      "step": 80
+    },
+    {
+      "entropy": 0.8265103206038475,
+      "epoch": 0.24140012070006034,
+      "grad_norm": 1.0106662511825562,
+      "learning_rate": 8.976478754121848e-05,
+      "loss": 0.7297060012817382,
+      "mean_token_accuracy": 0.7955138415098191,
+      "num_tokens": 236038.0,
+      "step": 100
+    },
+    {
+      "epoch": 0.24140012070006034,
+      "eval_entropy": 0.8122169536151244,
+      "eval_loss": 0.754906177520752,
+      "eval_mean_token_accuracy": 0.7917155255092664,
+      "eval_num_tokens": 236038.0,
+      "eval_runtime": 87.5754,
+      "eval_samples_per_second": 16.215,
+      "eval_steps_per_second": 2.033,
+      "step": 100
+    },
+    {
+      "entropy": 0.8098392844200134,
+      "epoch": 0.28968014484007243,
+      "grad_norm": 1.2238041162490845,
+      "learning_rate": 0.00010789908805459595,
+      "loss": 0.7201489448547364,
+      "mean_token_accuracy": 0.7975016921758652,
+      "num_tokens": 282918.0,
+      "step": 120
+    },
+    {
+      "epoch": 0.28968014484007243,
+      "eval_entropy": 0.8418726462326692,
+      "eval_loss": 0.7393178939819336,
+      "eval_mean_token_accuracy": 0.7936771585700217,
+      "eval_num_tokens": 282918.0,
+      "eval_runtime": 87.1597,
+      "eval_samples_per_second": 16.292,
+      "eval_steps_per_second": 2.042,
+      "step": 120
+    },
+    {
+      "entropy": 0.8080591082572937,
+      "epoch": 0.33796016898008446,
+      "grad_norm": 1.011591911315918,
+      "learning_rate": 0.00012603338856797343,
+      "loss": 0.7147697925567627,
+      "mean_token_accuracy": 0.7947175532579422,
+      "num_tokens": 331305.0,
+      "step": 140
+    },
+    {
+      "epoch": 0.33796016898008446,
+      "eval_entropy": 0.7894941431083037,
+      "eval_loss": 0.720219075679779,
+      "eval_mean_token_accuracy": 0.801899236239744,
+      "eval_num_tokens": 331305.0,
+      "eval_runtime": 87.1493,
+      "eval_samples_per_second": 16.294,
+      "eval_steps_per_second": 2.042,
+      "step": 140
+    },
+    {
+      "entropy": 0.7957205109298229,
+      "epoch": 0.38624019312009655,
+      "grad_norm": 0.8481830954551697,
+      "learning_rate": 0.00014416768908135088,
+      "loss": 0.7050027847290039,
+      "mean_token_accuracy": 0.7995368830859662,
+      "num_tokens": 375876.0,
+      "step": 160
+    },
+    {
+      "epoch": 0.38624019312009655,
+      "eval_entropy": 0.7746140591883928,
+      "eval_loss": 0.7175475358963013,
+      "eval_mean_token_accuracy": 0.7994848955213354,
+      "eval_num_tokens": 375876.0,
+      "eval_runtime": 86.655,
+      "eval_samples_per_second": 16.387,
+      "eval_steps_per_second": 2.054,
+      "step": 160
+    },
+    {
+      "entropy": 0.779565304517746,
+      "epoch": 0.43452021726010864,
+      "grad_norm": 1.032624363899231,
+      "learning_rate": 0.00016230198959472835,
+      "loss": 0.6978332042694092,
+      "mean_token_accuracy": 0.8015618294477462,
+      "num_tokens": 423913.0,
+      "step": 180
+    },
+    {
+      "epoch": 0.43452021726010864,
+      "eval_entropy": 0.7455583871080634,
+      "eval_loss": 0.7118850350379944,
+      "eval_mean_token_accuracy": 0.7999682195400923,
+      "eval_num_tokens": 423913.0,
+      "eval_runtime": 86.7842,
+      "eval_samples_per_second": 16.362,
+      "eval_steps_per_second": 2.051,
+      "step": 180
+    },
+    {
+      "entropy": 0.7810172818601131,
+      "epoch": 0.4828002414001207,
+      "grad_norm": 0.9276663064956665,
+      "learning_rate": 0.00018043629010810582,
+      "loss": 0.6937861442565918,
+      "mean_token_accuracy": 0.802893053740263,
+      "num_tokens": 472395.0,
+      "step": 200
+    },
+    {
+      "epoch": 0.4828002414001207,
+      "eval_entropy": 0.7650193290764027,
+      "eval_loss": 0.7020567059516907,
+      "eval_mean_token_accuracy": 0.8045302153973097,
+      "eval_num_tokens": 472395.0,
+      "eval_runtime": 88.326,
+      "eval_samples_per_second": 16.077,
+      "eval_steps_per_second": 2.015,
+      "step": 200
+    },
+    {
+      "entropy": 0.7862060204148292,
+      "epoch": 0.5310802655401328,
+      "grad_norm": 0.8195295929908752,
+      "learning_rate": 0.0001985705906214833,
+      "loss": 0.6878782272338867,
+      "mean_token_accuracy": 0.8027168907225132,
+      "num_tokens": 517049.0,
+      "step": 220
+    },
+    {
+      "epoch": 0.5310802655401328,
+      "eval_entropy": 0.7656565583154057,
+      "eval_loss": 0.6973932981491089,
+      "eval_mean_token_accuracy": 0.8047967278555538,
+      "eval_num_tokens": 517049.0,
+      "eval_runtime": 87.7698,
+      "eval_samples_per_second": 16.179,
+      "eval_steps_per_second": 2.028,
+      "step": 220
+    },
+    {
+      "entropy": 0.7611534461379051,
+      "epoch": 0.5793602896801449,
+      "grad_norm": 0.7499712109565735,
+      "learning_rate": 0.00021670489113486077,
+      "loss": 0.681937837600708,
+      "mean_token_accuracy": 0.8041978515684605,
+      "num_tokens": 562836.0,
+      "step": 240
+    },
+    {
+      "epoch": 0.5793602896801449,
+      "eval_entropy": 0.7599442380197933,
+      "eval_loss": 0.7002046704292297,
+      "eval_mean_token_accuracy": 0.8051861517884759,
+      "eval_num_tokens": 562836.0,
+      "eval_runtime": 87.9239,
+      "eval_samples_per_second": 16.15,
+      "eval_steps_per_second": 2.024,
+      "step": 240
+    },
+    {
+      "entropy": 0.7800783462822437,
+      "epoch": 0.627640313820157,
+      "grad_norm": 0.8645269274711609,
+      "learning_rate": 0.00023483919164823825,
+      "loss": 0.6954500675201416,
+      "mean_token_accuracy": 0.8031484372913837,
+      "num_tokens": 606612.0,
+      "step": 260
+    },
+    {
+      "epoch": 0.627640313820157,
+      "eval_entropy": 0.7498679090751691,
+      "eval_loss": 0.7046149373054504,
+      "eval_mean_token_accuracy": 0.802388215667746,
+      "eval_num_tokens": 606612.0,
+      "eval_runtime": 88.2468,
+      "eval_samples_per_second": 16.091,
+      "eval_steps_per_second": 2.017,
+      "step": 260
+    },
+    {
+      "entropy": 0.7586533665657044,
+      "epoch": 0.6759203379601689,
+      "grad_norm": 0.741596519947052,
+      "learning_rate": 0.0002529734921616157,
+      "loss": 0.690476369857788,
+      "mean_token_accuracy": 0.8046923100948333,
+      "num_tokens": 655920.0,
+      "step": 280
+    },
+    {
+      "epoch": 0.6759203379601689,
+      "eval_entropy": 0.7714274066887544,
+      "eval_loss": 0.6992160677909851,
+      "eval_mean_token_accuracy": 0.8042646482419432,
+      "eval_num_tokens": 655920.0,
+      "eval_runtime": 88.1754,
+      "eval_samples_per_second": 16.104,
+      "eval_steps_per_second": 2.019,
+      "step": 280
+    },
+    {
+      "entropy": 0.7585848286747933,
+      "epoch": 0.724200362100181,
+      "grad_norm": 1.0778124332427979,
+      "learning_rate": 0.00027110779267499317,
+      "loss": 0.6790409564971924,
+      "mean_token_accuracy": 0.8060351841151714,
+      "num_tokens": 702638.0,
+      "step": 300
+    },
+    {
+      "epoch": 0.724200362100181,
+      "eval_entropy": 0.7380402811457601,
+      "eval_loss": 0.7054820656776428,
+      "eval_mean_token_accuracy": 0.8003534738267406,
+      "eval_num_tokens": 702638.0,
+      "eval_runtime": 88.238,
+      "eval_samples_per_second": 16.093,
+      "eval_steps_per_second": 2.017,
+      "step": 300
+    },
+    {
+      "entropy": 0.7696981988847256,
+      "epoch": 0.7724803862401931,
+      "grad_norm": 0.8408658504486084,
+      "learning_rate": 0.00028924209318837064,
+      "loss": 0.7070387363433838,
+      "mean_token_accuracy": 0.8035947173833847,
+      "num_tokens": 749377.0,
+      "step": 320
+    },
+    {
+      "epoch": 0.7724803862401931,
+      "eval_entropy": 0.7897409023193831,
+      "eval_loss": 0.7015668749809265,
+      "eval_mean_token_accuracy": 0.805306687113944,
+      "eval_num_tokens": 749377.0,
+      "eval_runtime": 87.9395,
+      "eval_samples_per_second": 16.147,
+      "eval_steps_per_second": 2.024,
+      "step": 320
+    },
+    {
+      "entropy": 0.762337189912796,
+      "epoch": 0.8207604103802052,
+      "grad_norm": 0.9843310117721558,
+      "learning_rate": 0.0003073763937017481,
+      "loss": 0.7021088123321533,
+      "mean_token_accuracy": 0.8029616877436638,
+      "num_tokens": 798874.0,
+      "step": 340
+    },
+    {
+      "epoch": 0.8207604103802052,
+      "eval_entropy": 0.7210077121016685,
+      "eval_loss": 0.7082746624946594,
+      "eval_mean_token_accuracy": 0.8049283824609906,
+      "eval_num_tokens": 798874.0,
+      "eval_runtime": 88.1706,
+      "eval_samples_per_second": 16.105,
+      "eval_steps_per_second": 2.019,
+      "step": 340
+    },
+    {
+      "entropy": 0.7879025422036647,
+      "epoch": 0.8690404345202173,
+      "grad_norm": 0.9713814854621887,
+      "learning_rate": 0.0003255106942151256,
+      "loss": 0.714624547958374,
+      "mean_token_accuracy": 0.79987031519413,
+      "num_tokens": 840825.0,
+      "step": 360
+    },
+    {
+      "epoch": 0.8690404345202173,
+      "eval_entropy": 0.7866930489459735,
+      "eval_loss": 0.7145519852638245,
+      "eval_mean_token_accuracy": 0.8015657988157165,
+      "eval_num_tokens": 840825.0,
+      "eval_runtime": 88.3746,
+      "eval_samples_per_second": 16.068,
+      "eval_steps_per_second": 2.014,
+      "step": 360
+    },
+    {
+      "entropy": 0.7844726584851742,
+      "epoch": 0.9173204586602294,
+      "grad_norm": 1.2028311491012573,
+      "learning_rate": 0.00034364499472850306,
+      "loss": 0.7040786266326904,
+      "mean_token_accuracy": 0.7981885217130185,
+      "num_tokens": 882885.0,
+      "step": 380
+    },
+    {
+      "epoch": 0.9173204586602294,
+      "eval_entropy": 0.7767668339643585,
+      "eval_loss": 0.7073574066162109,
+      "eval_mean_token_accuracy": 0.8040851833445303,
+      "eval_num_tokens": 882885.0,
+      "eval_runtime": 88.3828,
+      "eval_samples_per_second": 16.066,
+      "eval_steps_per_second": 2.014,
+      "step": 380
+    },
+    {
+      "entropy": 0.7699790760874748,
+      "epoch": 0.9656004828002414,
+      "grad_norm": 0.9397398233413696,
+      "learning_rate": 0.00036177929524188054,
+      "loss": 0.7040313720703125,
+      "mean_token_accuracy": 0.8034727744758129,
+      "num_tokens": 927456.0,
+      "step": 400
+    },
+    {
+      "epoch": 0.9656004828002414,
+      "eval_entropy": 0.7749103545472863,
+      "eval_loss": 0.7093513011932373,
+      "eval_mean_token_accuracy": 0.8019482901926791,
+      "eval_num_tokens": 927456.0,
+      "eval_runtime": 86.2112,
+      "eval_samples_per_second": 16.471,
+      "eval_steps_per_second": 2.065,
+      "step": 400
+    },
+    {
+      "entropy": 0.7672389172888422,
+      "epoch": 1.012070006035003,
+      "grad_norm": 0.99210524559021,
+      "learning_rate": 0.00037628567078152296,
+      "loss": 0.6913118362426758,
+      "mean_token_accuracy": 0.8042210944287189,
+      "num_tokens": 972662.0,
+      "step": 420
+    },
+    {
+      "epoch": 1.012070006035003,
+      "eval_entropy": 0.7294771229283193,
+      "eval_loss": 0.7171520590782166,
+      "eval_mean_token_accuracy": 0.8015128585059991,
+      "eval_num_tokens": 972662.0,
+      "eval_runtime": 88.037,
+      "eval_samples_per_second": 16.13,
+      "eval_steps_per_second": 2.022,
+      "step": 420
+    },
+    {
+      "entropy": 0.7143943183124065,
+      "epoch": 1.060350030175015,
+      "grad_norm": 1.087638258934021,
+      "learning_rate": 0.0003762484015601069,
+      "loss": 0.642704439163208,
+      "mean_token_accuracy": 0.8124966286122799,
+      "num_tokens": 1022525.0,
+      "step": 440
+    },
+    {
+      "epoch": 1.060350030175015,
+      "eval_entropy": 0.743511886744017,
+      "eval_loss": 0.7116662859916687,
+      "eval_mean_token_accuracy": 0.8037210672758939,
+      "eval_num_tokens": 1022525.0,
+      "eval_runtime": 87.9931,
+      "eval_samples_per_second": 16.138,
+      "eval_steps_per_second": 2.023,
+      "step": 440
+    },
+    {
+      "entropy": 0.7237346574664116,
+      "epoch": 1.1086300543150271,
+      "grad_norm": 0.9365553259849548,
+      "learning_rate": 0.0003761579008392123,
+      "loss": 0.652644681930542,
+      "mean_token_accuracy": 0.8107496216893196,
+      "num_tokens": 1069194.0,
+      "step": 460
+    },
+    {
+      "epoch": 1.1086300543150271,
+      "eval_entropy": 0.7372823839776972,
+      "eval_loss": 0.7238303422927856,
+      "eval_mean_token_accuracy": 0.8026230124275336,
+      "eval_num_tokens": 1069194.0,
+      "eval_runtime": 87.6778,
+      "eval_samples_per_second": 16.196,
+      "eval_steps_per_second": 2.03,
+      "step": 460
+    },
+    {
+      "entropy": 0.7202137842774391,
+      "epoch": 1.1569100784550392,
+      "grad_norm": 0.7998443841934204,
+      "learning_rate": 0.0003760141942294726,
+      "loss": 0.6485219478607178,
+      "mean_token_accuracy": 0.8092714451253414,
+      "num_tokens": 1117570.0,
+      "step": 480
+    },
+    {
+      "epoch": 1.1569100784550392,
+      "eval_entropy": 0.71547647942318,
+      "eval_loss": 0.7124277949333191,
+      "eval_mean_token_accuracy": 0.8013174219077892,
+      "eval_num_tokens": 1117570.0,
+      "eval_runtime": 87.4387,
+      "eval_samples_per_second": 16.24,
+      "eval_steps_per_second": 2.036,
+      "step": 480
+    },
+    {
+      "entropy": 0.7270819500088692,
+      "epoch": 1.2051901025950513,
+      "grad_norm": 0.9470787644386292,
+      "learning_rate": 0.00037581732239815854,
+      "loss": 0.66352219581604,
+      "mean_token_accuracy": 0.8102593503892421,
+      "num_tokens": 1161717.0,
+      "step": 500
+    },
+    {
+      "epoch": 1.2051901025950513,
+      "eval_entropy": 0.7791565897759427,
+      "eval_loss": 0.7179726958274841,
+      "eval_mean_token_accuracy": 0.7948898328154275,
+      "eval_num_tokens": 1161717.0,
+      "eval_runtime": 87.9927,
+      "eval_samples_per_second": 16.138,
+      "eval_steps_per_second": 2.023,
+      "step": 500
+    },
+    {
+      "entropy": 0.7320936664938926,
+      "epoch": 1.2534701267350634,
+      "grad_norm": 1.0220379829406738,
+      "learning_rate": 0.0003755673410576695,
+      "loss": 0.6581549644470215,
+      "mean_token_accuracy": 0.8098709337413311,
+      "num_tokens": 1209390.0,
+      "step": 520
+    },
+    {
+      "epoch": 1.2534701267350634,
+      "eval_entropy": 0.7710678225822662,
+      "eval_loss": 0.708797812461853,
+      "eval_mean_token_accuracy": 0.8039572778042783,
+      "eval_num_tokens": 1209390.0,
+      "eval_runtime": 88.1193,
+      "eval_samples_per_second": 16.115,
+      "eval_steps_per_second": 2.02,
+      "step": 520
+    },
+    {
+      "entropy": 0.7281674664467573,
+      "epoch": 1.3017501508750755,
+      "grad_norm": 0.8775188326835632,
+      "learning_rate": 0.000375264320949768,
+      "loss": 0.6631278991699219,
+      "mean_token_accuracy": 0.8101261422038079,
+      "num_tokens": 1256611.0,
+      "step": 540
+    },
+    {
+      "epoch": 1.3017501508750755,
+      "eval_entropy": 0.7199799800187014,
+      "eval_loss": 0.6981531381607056,
+      "eval_mean_token_accuracy": 0.8071265254127845,
+      "eval_num_tokens": 1256611.0,
+      "eval_runtime": 88.165,
+      "eval_samples_per_second": 16.106,
+      "eval_steps_per_second": 2.019,
+      "step": 540
+    },
+    {
+      "entropy": 0.722860773652792,
+      "epoch": 1.3500301750150876,
+      "grad_norm": 0.8949288129806519,
+      "learning_rate": 0.00037490834782556,
+      "loss": 0.6639598846435547,
+      "mean_token_accuracy": 0.8105649061501026,
+      "num_tokens": 1302325.0,
+      "step": 560
+    },
+    {
+      "epoch": 1.3500301750150876,
+      "eval_entropy": 0.7134760032209118,
+      "eval_loss": 0.7029948234558105,
+      "eval_mean_token_accuracy": 0.8062985880991046,
+      "eval_num_tokens": 1302325.0,
+      "eval_runtime": 88.1286,
+      "eval_samples_per_second": 16.113,
+      "eval_steps_per_second": 2.02,
+      "step": 560
+    },
+    {
+      "entropy": 0.7299126699566841,
+      "epoch": 1.3983101991550995,
+      "grad_norm": 1.0323039293289185,
+      "learning_rate": 0.0003744995224212291,
+      "loss": 0.6535763263702392,
+      "mean_token_accuracy": 0.8120224848389626,
+      "num_tokens": 1343844.0,
+      "step": 580
+    },
+    {
+      "epoch": 1.3983101991550995,
+      "eval_entropy": 0.7169701182440426,
+      "eval_loss": 0.7067857980728149,
+      "eval_mean_token_accuracy": 0.8040432936689826,
+      "eval_num_tokens": 1343844.0,
+      "eval_runtime": 88.2252,
+      "eval_samples_per_second": 16.095,
+      "eval_steps_per_second": 2.018,
+      "step": 580
+    },
+    {
+      "entropy": 0.7281742256134749,
+      "epoch": 1.4465902232951118,
+      "grad_norm": 1.3370424509048462,
+      "learning_rate": 0.00037403796042952863,
+      "loss": 0.6637272357940673,
+      "mean_token_accuracy": 0.8118261776864528,
+      "num_tokens": 1389471.0,
+      "step": 600
+    },
+    {
+      "epoch": 1.4465902232951118,
+      "eval_entropy": 0.7378843212395572,
+      "eval_loss": 0.7038390040397644,
+      "eval_mean_token_accuracy": 0.8059229371922739,
+      "eval_num_tokens": 1389471.0,
+      "eval_runtime": 87.9301,
+      "eval_samples_per_second": 16.149,
+      "eval_steps_per_second": 2.024,
+      "step": 600
+    },
+    {
+      "entropy": 0.7388614989817143,
+      "epoch": 1.4948702474351236,
+      "grad_norm": 0.8464171290397644,
+      "learning_rate": 0.00037352379246704257,
+      "loss": 0.6604740142822265,
+      "mean_token_accuracy": 0.8102974124252796,
+      "num_tokens": 1435975.0,
+      "step": 620
+    },
+    {
+      "epoch": 1.4948702474351236,
+      "eval_entropy": 0.7126847046814607,
+      "eval_loss": 0.6950703263282776,
+      "eval_mean_token_accuracy": 0.8085885392815879,
+      "eval_num_tokens": 1435975.0,
+      "eval_runtime": 88.2732,
+      "eval_samples_per_second": 16.086,
+      "eval_steps_per_second": 2.016,
+      "step": 620
+    },
+    {
+      "entropy": 0.7258916139602661,
+      "epoch": 1.5431502715751357,
+      "grad_norm": 1.1780085563659668,
+      "learning_rate": 0.0003729571640372223,
+      "loss": 0.6529891014099121,
+      "mean_token_accuracy": 0.8085981778800487,
+      "num_tokens": 1482035.0,
+      "step": 640
+    },
+    {
+      "epoch": 1.5431502715751357,
+      "eval_entropy": 0.7096509638797032,
+      "eval_loss": 0.690051257610321,
+      "eval_mean_token_accuracy": 0.8100520241796301,
+      "eval_num_tokens": 1482035.0,
+      "eval_runtime": 88.3153,
+      "eval_samples_per_second": 16.079,
+      "eval_steps_per_second": 2.016,
+      "step": 640
+    },
+    {
+      "entropy": 0.7151173129677773,
+      "epoch": 1.5914302957151478,
+      "grad_norm": 1.0357667207717896,
+      "learning_rate": 0.0003723382354892108,
+      "loss": 0.6451474189758301,
+      "mean_token_accuracy": 0.8136902332305909,
+      "num_tokens": 1528119.0,
+      "step": 660
+    },
+    {
+      "epoch": 1.5914302957151478,
+      "eval_entropy": 0.7060252652409371,
+      "eval_loss": 0.6899483799934387,
+      "eval_mean_token_accuracy": 0.810834194502134,
+      "eval_num_tokens": 1528119.0,
+      "eval_runtime": 87.8848,
+      "eval_samples_per_second": 16.158,
+      "eval_steps_per_second": 2.025,
+      "step": 660
+    },
+    {
+      "entropy": 0.7141377851366997,
+      "epoch": 1.63971031985516,
+      "grad_norm": 0.8391667008399963,
+      "learning_rate": 0.000371667181972466,
+      "loss": 0.6481579780578614,
+      "mean_token_accuracy": 0.8159015104174614,
+      "num_tokens": 1573661.0,
+      "step": 680
+    },
+    {
+      "epoch": 1.63971031985516,
+      "eval_entropy": 0.7598993319473909,
+      "eval_loss": 0.681204080581665,
+      "eval_mean_token_accuracy": 0.8100194586126992,
+      "eval_num_tokens": 1573661.0,
+      "eval_runtime": 88.2415,
+      "eval_samples_per_second": 16.092,
+      "eval_steps_per_second": 2.017,
+      "step": 680
+    },
+    {
+      "entropy": 0.7235980287194252,
+      "epoch": 1.687990343995172,
+      "grad_norm": 0.8197196125984192,
+      "learning_rate": 0.00037094419338719537,
+      "loss": 0.6582849025726318,
+      "mean_token_accuracy": 0.8114834539592266,
+      "num_tokens": 1621714.0,
+      "step": 700
+    },
+    {
+      "epoch": 1.687990343995172,
+      "eval_entropy": 0.7166854568411795,
+      "eval_loss": 0.679177463054657,
+      "eval_mean_token_accuracy": 0.8119039877077167,
+      "eval_num_tokens": 1621714.0,
+      "eval_runtime": 88.0957,
+      "eval_samples_per_second": 16.119,
+      "eval_steps_per_second": 2.021,
+      "step": 700
+    },
+    {
+      "entropy": 0.7278190158307553,
+      "epoch": 1.736270368135184,
+      "grad_norm": 0.7820602655410767,
+      "learning_rate": 0.0003701694743306164,
+      "loss": 0.6633153438568116,
+      "mean_token_accuracy": 0.8123249113559723,
+      "num_tokens": 1672118.0,
+      "step": 720
+    },
+    {
+      "epoch": 1.736270368135184,
+      "eval_entropy": 0.6980850314826108,
+      "eval_loss": 0.6775653958320618,
+      "eval_mean_token_accuracy": 0.8116872906684875,
+      "eval_num_tokens": 1672118.0,
+      "eval_runtime": 87.9795,
+      "eval_samples_per_second": 16.14,
+      "eval_steps_per_second": 2.023,
+      "step": 720
+    },
+    {
+      "entropy": 0.7149631556123495,
+      "epoch": 1.7845503922751962,
+      "grad_norm": 1.0139760971069336,
+      "learning_rate": 0.00036934324403905795,
+      "loss": 0.6468976020812989,
+      "mean_token_accuracy": 0.8117561548948288,
+      "num_tokens": 1720465.0,
+      "step": 740
+    },
+    {
+      "epoch": 1.7845503922751962,
+      "eval_entropy": 0.69009849094273,
+      "eval_loss": 0.6731093525886536,
+      "eval_mean_token_accuracy": 0.81281964196248,
+      "eval_num_tokens": 1720465.0,
+      "eval_runtime": 87.7412,
+      "eval_samples_per_second": 16.184,
+      "eval_steps_per_second": 2.029,
+      "step": 740
+    },
+    {
+      "entropy": 0.7226049453020096,
+      "epoch": 1.832830416415208,
+      "grad_norm": 1.5040546655654907,
+      "learning_rate": 0.0003684657363259193,
+      "loss": 0.6385328769683838,
+      "mean_token_accuracy": 0.8132639616727829,
+      "num_tokens": 1764440.0,
+      "step": 760
+    },
+    {
+      "epoch": 1.832830416415208,
+      "eval_entropy": 0.7125169618076153,
+      "eval_loss": 0.6864680647850037,
+      "eval_mean_token_accuracy": 0.8049418089095126,
+      "eval_num_tokens": 1764440.0,
+      "eval_runtime": 87.9429,
+      "eval_samples_per_second": 16.147,
+      "eval_steps_per_second": 2.024,
+      "step": 760
+    },
+    {
+      "entropy": 0.707821810990572,
+      "epoch": 1.8811104405552204,
+      "grad_norm": 0.8604223132133484,
+      "learning_rate": 0.00036753719951550327,
+      "loss": 0.6397049427032471,
+      "mean_token_accuracy": 0.8137877315282822,
+      "num_tokens": 1810828.0,
+      "step": 780
+    },
+    {
+      "epoch": 1.8811104405552204,
+      "eval_entropy": 0.6492102551326323,
+      "eval_loss": 0.6743726134300232,
+      "eval_mean_token_accuracy": 0.8130356572317273,
+      "eval_num_tokens": 1810828.0,
+      "eval_runtime": 87.8684,
+      "eval_samples_per_second": 16.161,
+      "eval_steps_per_second": 2.026,
+      "step": 780
+    },
+    {
+      "entropy": 0.6800355084240437,
+      "epoch": 1.9293904646952322,
+      "grad_norm": 0.7270982265472412,
+      "learning_rate": 0.00036655789637274377,
+      "loss": 0.6267755031585693,
+      "mean_token_accuracy": 0.8180945307016373,
+      "num_tokens": 1857600.0,
+      "step": 800
+    },
+    {
+      "epoch": 1.9293904646952322,
+      "eval_entropy": 0.6727538651294922,
+      "eval_loss": 0.6696028709411621,
+      "eval_mean_token_accuracy": 0.8116443090224534,
+      "eval_num_tokens": 1857600.0,
+      "eval_runtime": 88.2449,
+      "eval_samples_per_second": 16.092,
+      "eval_steps_per_second": 2.017,
+      "step": 800
+    },
+    {
+      "entropy": 0.6746691606938839,
+      "epoch": 1.9776704888352445,
+      "grad_norm": 0.9771540760993958,
+      "learning_rate": 0.0003655281040288461,
+      "loss": 0.6326992988586426,
+      "mean_token_accuracy": 0.8179976396262646,
+      "num_tokens": 1901998.0,
+      "step": 820
+    },
+    {
+      "epoch": 1.9776704888352445,
+      "eval_entropy": 0.6636319716324967,
+      "eval_loss": 0.6714363098144531,
+      "eval_mean_token_accuracy": 0.8126635943235976,
+      "eval_num_tokens": 1901998.0,
+      "eval_runtime": 87.7666,
+      "eval_samples_per_second": 16.179,
+      "eval_steps_per_second": 2.028,
+      "step": 820
+    },
+    {
+      "entropy": 0.6319125778489298,
+      "epoch": 2.024140012070006,
+      "grad_norm": 0.8380444645881653,
+      "learning_rate": 0.0003644481139028622,
+      "loss": 0.5780903816223144,
+      "mean_token_accuracy": 0.8250188401767186,
+      "num_tokens": 1947553.0,
+      "step": 840
+    },
+    {
+      "epoch": 2.024140012070006,
+      "eval_entropy": 0.6627151989535,
+      "eval_loss": 0.6774935722351074,
+      "eval_mean_token_accuracy": 0.8140510819601209,
+      "eval_num_tokens": 1947553.0,
+      "eval_runtime": 87.8287,
+      "eval_samples_per_second": 16.168,
+      "eval_steps_per_second": 2.027,
+      "step": 840
+    },
+    {
+      "entropy": 0.6059147048741579,
+      "epoch": 2.0724200362100182,
+      "grad_norm": 0.9745152592658997,
+      "learning_rate": 0.0003633182316192229,
+      "loss": 0.546702527999878,
+      "mean_token_accuracy": 0.8358186542987823,
+      "num_tokens": 1995234.0,
+      "step": 860
+    },
+    {
+      "epoch": 2.0724200362100182,
+      "eval_entropy": 0.6620710384979677,
+      "eval_loss": 0.6769291162490845,
+      "eval_mean_token_accuracy": 0.8135082835561773,
+      "eval_num_tokens": 1995234.0,
+      "eval_runtime": 88.0008,
+      "eval_samples_per_second": 16.136,
+      "eval_steps_per_second": 2.023,
+      "step": 860
+    },
+    {
+      "entropy": 0.6076117843389511,
+      "epoch": 2.12070006035003,
+      "grad_norm": 0.6914380192756653,
+      "learning_rate": 0.0003621387769212491,
+      "loss": 0.5324535846710206,
+      "mean_token_accuracy": 0.8366568259894848,
+      "num_tokens": 2040639.0,
+      "step": 880
+    },
+    {
+      "epoch": 2.12070006035003,
+      "eval_entropy": 0.6592224213514435,
+      "eval_loss": 0.670383632183075,
+      "eval_mean_token_accuracy": 0.8162150088320957,
+      "eval_num_tokens": 2040639.0,
+      "eval_runtime": 88.324,
+      "eval_samples_per_second": 16.077,
+      "eval_steps_per_second": 2.015,
+      "step": 880
+    },
+    {
+      "entropy": 0.6158072650432587,
+      "epoch": 2.1689800844900424,
+      "grad_norm": 0.9649198055267334,
+      "learning_rate": 0.0003609100835806688,
+      "loss": 0.5564568996429443,
+      "mean_token_accuracy": 0.8329654954373836,
+      "num_tokens": 2090604.0,
+      "step": 900
+    },
+    {
+      "epoch": 2.1689800844900424,
+      "eval_entropy": 0.608588819758276,
+      "eval_loss": 0.6745102405548096,
+      "eval_mean_token_accuracy": 0.8155080542135774,
+      "eval_num_tokens": 2090604.0,
+      "eval_runtime": 87.52,
+      "eval_samples_per_second": 16.225,
+      "eval_steps_per_second": 2.034,
+      "step": 900
+    },
+    {
+      "entropy": 0.5938734326511621,
+      "epoch": 2.2172601086300543,
+      "grad_norm": 0.6915557384490967,
+      "learning_rate": 0.0003596324993031632,
+      "loss": 0.5351875782012939,
+      "mean_token_accuracy": 0.8385631121695042,
+      "num_tokens": 2142246.0,
+      "step": 920
+    },
+    {
+      "epoch": 2.2172601086300543,
+      "eval_entropy": 0.6251631205001574,
+      "eval_loss": 0.6664518117904663,
+      "eval_mean_token_accuracy": 0.81711940417129,
+      "eval_num_tokens": 2142246.0,
+      "eval_runtime": 87.1792,
+      "eval_samples_per_second": 16.288,
+      "eval_steps_per_second": 2.042,
+      "step": 920
+    },
+    {
+      "entropy": 0.623173613846302,
+      "epoch": 2.2655401327700666,
+      "grad_norm": 0.8366610407829285,
+      "learning_rate": 0.00035830638562997063,
+      "loss": 0.5488213539123535,
+      "mean_token_accuracy": 0.8321483485400677,
+      "num_tokens": 2183144.0,
+      "step": 940
+    },
+    {
+      "epoch": 2.2655401327700666,
+      "eval_entropy": 0.6237085649136747,
+      "eval_loss": 0.68075031042099,
+      "eval_mean_token_accuracy": 0.814671738429016,
+      "eval_num_tokens": 2183144.0,
+      "eval_runtime": 88.0652,
+      "eval_samples_per_second": 16.124,
+      "eval_steps_per_second": 2.021,
+      "step": 940
+    },
+    {
+      "entropy": 0.6236911740154027,
+      "epoch": 2.3138201569100785,
+      "grad_norm": 0.7812452912330627,
+      "learning_rate": 0.00035693211783557416,
+      "loss": 0.5696200847625732,
+      "mean_token_accuracy": 0.82991396561265,
+      "num_tokens": 2228098.0,
+      "step": 960
+    },
+    {
+      "epoch": 2.3138201569100785,
+      "eval_entropy": 0.6327033426319615,
+      "eval_loss": 0.6742984652519226,
+      "eval_mean_token_accuracy": 0.8148086091105857,
+      "eval_num_tokens": 2228098.0,
+      "eval_runtime": 88.1711,
+      "eval_samples_per_second": 16.105,
+      "eval_steps_per_second": 2.019,
+      "step": 960
+    },
+    {
+      "entropy": 0.6218426413834095,
+      "epoch": 2.3621001810500903,
+      "grad_norm": 0.8719790577888489,
+      "learning_rate": 0.0003555100848215035,
+      "loss": 0.5582651615142822,
+      "mean_token_accuracy": 0.8323961742222309,
+      "num_tokens": 2273062.0,
+      "step": 980
+    },
+    {
+      "epoch": 2.3621001810500903,
+      "eval_entropy": 0.6338012375858393,
+      "eval_loss": 0.6685822606086731,
+      "eval_mean_token_accuracy": 0.8152180300669724,
+      "eval_num_tokens": 2273062.0,
+      "eval_runtime": 88.101,
+      "eval_samples_per_second": 16.118,
+      "eval_steps_per_second": 2.02,
+      "step": 980
+    },
+    {
+      "entropy": 0.6050681818276644,
+      "epoch": 2.4103802051901027,
+      "grad_norm": 0.9151566028594971,
+      "learning_rate": 0.00035404068900628076,
+      "loss": 0.559452486038208,
+      "mean_token_accuracy": 0.8310242861509323,
+      "num_tokens": 2319305.0,
+      "step": 1000
+    },
+    {
+      "epoch": 2.4103802051901027,
+      "eval_entropy": 0.6634890969549672,
+      "eval_loss": 0.6692465543746948,
+      "eval_mean_token_accuracy": 0.8134353673190213,
+      "eval_num_tokens": 2319305.0,
+      "eval_runtime": 87.7711,
+      "eval_samples_per_second": 16.178,
+      "eval_steps_per_second": 2.028,
+      "step": 1000
+    },
+    {
+      "entropy": 0.6169968567788601,
+      "epoch": 2.4586602293301145,
+      "grad_norm": 0.7081006765365601,
+      "learning_rate": 0.0003525243462115402,
+      "loss": 0.5589694023132324,
+      "mean_token_accuracy": 0.8357502184808254,
+      "num_tokens": 2368592.0,
+      "step": 1020
+    },
+    {
+      "epoch": 2.4586602293301145,
+      "eval_entropy": 0.6126393578695447,
+      "eval_loss": 0.6706712245941162,
+      "eval_mean_token_accuracy": 0.8155577500884452,
+      "eval_num_tokens": 2368592.0,
+      "eval_runtime": 88.1407,
+      "eval_samples_per_second": 16.111,
+      "eval_steps_per_second": 2.019,
+      "step": 1020
+    },
+    {
+      "entropy": 0.6072248566895724,
+      "epoch": 2.506940253470127,
+      "grad_norm": 0.6999370455741882,
+      "learning_rate": 0.0003509614855443561,
+      "loss": 0.5500593185424805,
+      "mean_token_accuracy": 0.8337264291942119,
+      "num_tokens": 2412859.0,
+      "step": 1040
+    },
+    {
+      "epoch": 2.506940253470127,
+      "eval_entropy": 0.6417417074187418,
+      "eval_loss": 0.6651163697242737,
+      "eval_mean_token_accuracy": 0.8159113908081912,
+      "eval_num_tokens": 2412859.0,
+      "eval_runtime": 87.4854,
+      "eval_samples_per_second": 16.231,
+      "eval_steps_per_second": 2.035,
+      "step": 1040
+    },
+    {
+      "entropy": 0.6301592070609331,
+      "epoch": 2.5552202776101387,
+      "grad_norm": 0.8768910765647888,
+      "learning_rate": 0.00034935254927581064,
+      "loss": 0.5613903999328613,
+      "mean_token_accuracy": 0.8298896946012974,
+      "num_tokens": 2458201.0,
+      "step": 1060
+    },
+    {
+      "epoch": 2.5552202776101387,
+      "eval_entropy": 0.6569794751285167,
+      "eval_loss": 0.6668341159820557,
+      "eval_mean_token_accuracy": 0.8126791795987761,
+      "eval_num_tokens": 2458201.0,
+      "eval_runtime": 88.0474,
+      "eval_samples_per_second": 16.128,
+      "eval_steps_per_second": 2.022,
+      "step": 1060
+    }
+  ],
+  "logging_steps": 20,
+  "max_steps": 4150,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 10,
+  "save_steps": 20,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 1.0385749388429107e+17,
+  "train_batch_size": 4,
+  "trial_name": null,
+  "trial_params": null
+}

overgeneralisation_original_Estonian/Qwen3.5-4B-Base_overgeneralisation_splits_original_features_train_overgeneralisation_splits_original_features_test2/checkpoint-1080/README.md

@@ -0,0 +1,209 @@
+---
+base_model: Qwen/Qwen3.5-4B-Base
+library_name: peft
+pipeline_tag: text-generation
+tags:
+- base_model:adapter:Qwen/Qwen3.5-4B-Base
+- lora
+- sft
+- transformers
+- trl
+---
+
+# Model Card for Model ID
+
+<!-- Provide a quick summary of what the model is/does. -->
+
+
+
+## Model Details
+
+### Model Description
+
+<!-- Provide a longer summary of what this model is. -->
+
+
+
+- **Developed by:** [More Information Needed]
+- **Funded by [optional]:** [More Information Needed]
+- **Shared by [optional]:** [More Information Needed]
+- **Model type:** [More Information Needed]
+- **Language(s) (NLP):** [More Information Needed]
+- **License:** [More Information Needed]
+- **Finetuned from model [optional]:** [More Information Needed]
+
+### Model Sources [optional]
+
+<!-- Provide the basic links for the model. -->
+
+- **Repository:** [More Information Needed]
+- **Paper [optional]:** [More Information Needed]
+- **Demo [optional]:** [More Information Needed]
+
+## Uses
+
+<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
+
+### Direct Use
+
+<!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
+
+[More Information Needed]
+
+### Downstream Use [optional]
+
+<!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->
+
+[More Information Needed]
+
+### Out-of-Scope Use
+
+<!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
+
+[More Information Needed]
+
+## Bias, Risks, and Limitations
+
+<!-- This section is meant to convey both technical and sociotechnical limitations. -->
+
+[More Information Needed]
+
+### Recommendations
+
+<!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->
+
+Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
+
+## How to Get Started with the Model
+
+Use the code below to get started with the model.
+
+[More Information Needed]
+
+## Training Details
+
+### Training Data
+
+<!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
+
+[More Information Needed]
+
+### Training Procedure
+
+<!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
+
+#### Preprocessing [optional]
+
+[More Information Needed]
+
+
+#### Training Hyperparameters
+
+- **Training regime:** [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->
+
+#### Speeds, Sizes, Times [optional]
+
+<!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
+
+[More Information Needed]
+
+## Evaluation
+
+<!-- This section describes the evaluation protocols and provides the results. -->
+
+### Testing Data, Factors & Metrics
+
+#### Testing Data
+
+<!-- This should link to a Dataset Card if possible. -->
+
+[More Information Needed]
+
+#### Factors
+
+<!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->
+
+[More Information Needed]
+
+#### Metrics
+
+<!-- These are the evaluation metrics being used, ideally with a description of why. -->
+
+[More Information Needed]
+
+### Results
+
+[More Information Needed]
+
+#### Summary
+
+
+
+## Model Examination [optional]
+
+<!-- Relevant interpretability work for the model goes here -->
+
+[More Information Needed]
+
+## Environmental Impact
+
+<!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly -->
+
+Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700).
+
+- **Hardware Type:** [More Information Needed]
+- **Hours used:** [More Information Needed]
+- **Cloud Provider:** [More Information Needed]
+- **Compute Region:** [More Information Needed]
+- **Carbon Emitted:** [More Information Needed]
+
+## Technical Specifications [optional]
+
+### Model Architecture and Objective
+
+[More Information Needed]
+
+### Compute Infrastructure
+
+[More Information Needed]
+
+#### Hardware
+
+[More Information Needed]
+
+#### Software
+
+[More Information Needed]
+
+## Citation [optional]
+
+<!-- If there is a paper or blog post introducing the model, the APA and Bibtex information for that should go in this section. -->
+
+**BibTeX:**
+
+[More Information Needed]
+
+**APA:**
+
+[More Information Needed]
+
+## Glossary [optional]
+
+<!-- If relevant, include terms and calculations in this section that can help readers understand the model or model card. -->
+
+[More Information Needed]
+
+## More Information [optional]
+
+[More Information Needed]
+
+## Model Card Authors [optional]
+
+[More Information Needed]
+
+## Model Card Contact
+
+[More Information Needed]
+### Framework versions
+
+- PEFT 0.18.1

overgeneralisation_original_Estonian/Qwen3.5-4B-Base_overgeneralisation_splits_original_features_train_overgeneralisation_splits_original_features_test2/checkpoint-1080/adapter_config.json

@@ -0,0 +1,46 @@
+{
+  "alora_invocation_tokens": null,
+  "alpha_pattern": {},
+  "arrow_config": null,
+  "auto_mapping": null,
+  "base_model_name_or_path": "Qwen/Qwen3.5-4B-Base",
+  "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": 128,
+  "lora_bias": false,
+  "lora_dropout": 0.03828026524568501,
+  "megatron_config": null,
+  "megatron_core": "megatron.core",
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "peft_version": "0.18.1",
+  "qalora_group_size": 16,
+  "r": 64,
+  "rank_pattern": {},
+  "revision": null,
+  "target_modules": [
+    "gate_proj",
+    "v_proj",
+    "up_proj",
+    "down_proj",
+    "o_proj",
+    "q_proj",
+    "k_proj"
+  ],
+  "target_parameters": null,
+  "task_type": "CAUSAL_LM",
+  "trainable_token_indices": null,
+  "use_dora": false,
+  "use_qalora": false,
+  "use_rslora": false
+}

overgeneralisation_original_Estonian/Qwen3.5-4B-Base_overgeneralisation_splits_original_features_train_overgeneralisation_splits_original_features_test2/checkpoint-1080/chat_template.jinja

@@ -0,0 +1,154 @@
+{%- set image_count = namespace(value=0) %}
+{%- set video_count = namespace(value=0) %}
+{%- macro render_content(content, do_vision_count, is_system_content=false) %}
+    {%- if content is string %}
+        {{- content }}
+    {%- elif content is iterable and content is not mapping %}
+        {%- for item in content %}
+            {%- if 'image' in item or 'image_url' in item or item.type == 'image' %}
+                {%- if is_system_content %}
+                    {{- raise_exception('System message cannot contain images.') }}
+                {%- endif %}
+                {%- if do_vision_count %}
+                    {%- set image_count.value = image_count.value + 1 %}
+                {%- endif %}
+                {%- if add_vision_id %}
+                    {{- 'Picture ' ~ image_count.value ~ ': ' }}
+                {%- endif %}
+                {{- '<|vision_start|><|image_pad|><|vision_end|>' }}
+            {%- elif 'video' in item or item.type == 'video' %}
+                {%- if is_system_content %}
+                    {{- raise_exception('System message cannot contain videos.') }}
+                {%- endif %}
+                {%- if do_vision_count %}
+                    {%- set video_count.value = video_count.value + 1 %}
+                {%- endif %}
+                {%- if add_vision_id %}
+                    {{- 'Video ' ~ video_count.value ~ ': ' }}
+                {%- endif %}
+                {{- '<|vision_start|><|video_pad|><|vision_end|>' }}
+            {%- elif 'text' in item %}
+                {{- item.text }}
+            {%- else %}
+                {{- raise_exception('Unexpected item type in content.') }}
+            {%- endif %}
+        {%- endfor %}
+    {%- elif content is none or content is undefined %}
+        {{- '' }}
+    {%- else %}
+        {{- raise_exception('Unexpected content type.') }}
+    {%- endif %}
+{%- endmacro %}
+{%- if not messages %}
+    {{- raise_exception('No messages provided.') }}
+{%- endif %}
+{%- if tools and tools is iterable and tools is not mapping %}
+    {{- '<|im_start|>system\n' }}
+    {{- "# Tools\n\nYou have access to the following functions:\n\n<tools>" }}
+    {%- for tool in tools %}
+        {{- "\n" }}
+        {{- tool | tojson }}
+    {%- endfor %}
+    {{- "\n</tools>" }}
+    {{- '\n\nIf you choose to call a function ONLY reply in the following format with NO suffix:\n\n<tool_call>\n<function=example_function_name>\n<parameter=example_parameter_1>\nvalue_1\n</parameter>\n<parameter=example_parameter_2>\nThis is the value for the second parameter\nthat can span\nmultiple lines\n</parameter>\n</function>\n</tool_call>\n\n<IMPORTANT>\nReminder:\n- Function calls MUST follow the specified format: an inner <function=...></function> block must be nested within <tool_call></tool_call> XML tags\n- Required parameters MUST be specified\n- You may provide optional reasoning for your function call in natural language BEFORE the function call, but NOT after\n- If there is no function call available, answer the question like normal with your current knowledge and do not tell the user about function calls\n</IMPORTANT>' }}
+    {%- if messages[0].role == 'system' %}
+        {%- set content = render_content(messages[0].content, false, true)|trim %}
+        {%- if content %}
+            {{- '\n\n' + content }}
+        {%- endif %}
+    {%- endif %}
+    {{- '<|im_end|>\n' }}
+{%- else %}
+    {%- if messages[0].role == 'system' %}
+        {%- set content = render_content(messages[0].content, false, true)|trim %}
+        {{- '<|im_start|>system\n' + content + '<|im_end|>\n' }}
+    {%- endif %}
+{%- endif %}
+{%- set ns = namespace(multi_step_tool=true, last_query_index=messages|length - 1) %}
+{%- for message in messages[::-1] %}
+    {%- set index = (messages|length - 1) - loop.index0 %}
+    {%- if ns.multi_step_tool and message.role == "user" %}
+        {%- set content = render_content(message.content, false)|trim %}
+        {%- if not(content.startswith('<tool_response>') and content.endswith('</tool_response>')) %}
+            {%- set ns.multi_step_tool = false %}
+            {%- set ns.last_query_index = index %}
+        {%- endif %}
+    {%- endif %}
+{%- endfor %}
+{%- if ns.multi_step_tool %}
+    {{- raise_exception('No user query found in messages.') }}
+{%- endif %}
+{%- for message in messages %}
+    {%- set content = render_content(message.content, true)|trim %}
+    {%- if message.role == "system" %}
+        {%- if not loop.first %}
+            {{- raise_exception('System message must be at the beginning.') }}
+        {%- endif %}
+    {%- elif message.role == "user" %}
+        {{- '<|im_start|>' + message.role + '\n' + content + '<|im_end|>' + '\n' }}
+    {%- elif message.role == "assistant" %}
+        {%- set reasoning_content = '' %}
+        {%- if message.reasoning_content is string %}
+            {%- set reasoning_content = message.reasoning_content %}
+        {%- else %}
+            {%- if '</think>' in content %}
+                {%- set reasoning_content = content.split('</think>')[0].rstrip('\n').split('<think>')[-1].lstrip('\n') %}
+                {%- set content = content.split('</think>')[-1].lstrip('\n') %}
+            {%- endif %}
+        {%- endif %}
+        {%- set reasoning_content = reasoning_content|trim %}
+        {%- if loop.index0 > ns.last_query_index %}
+            {{- '<|im_start|>' + message.role + '\n<think>\n' + reasoning_content + '\n</think>\n\n' + content }}
+        {%- else %}
+            {{- '<|im_start|>' + message.role + '\n' + content }}
+        {%- endif %}
+        {%- if message.tool_calls and message.tool_calls is iterable and message.tool_calls is not mapping %}
+            {%- for tool_call in message.tool_calls %}
+                {%- if tool_call.function is defined %}
+                    {%- set tool_call = tool_call.function %}
+                {%- endif %}
+                {%- if loop.first %}
+                    {%- if content|trim %}
+                        {{- '\n\n<tool_call>\n<function=' + tool_call.name + '>\n' }}
+                    {%- else %}
+                        {{- '<tool_call>\n<function=' + tool_call.name + '>\n' }}
+                    {%- endif %}
+                {%- else %}
+                    {{- '\n<tool_call>\n<function=' + tool_call.name + '>\n' }}
+                {%- endif %}
+                {%- if tool_call.arguments is defined %}
+                    {%- for args_name, args_value in tool_call.arguments|items %}
+                        {{- '<parameter=' + args_name + '>\n' }}
+                        {%- set args_value = args_value | tojson | safe if args_value is mapping or (args_value is sequence and args_value is not string) else args_value | string %}
+                        {{- args_value }}
+                        {{- '\n</parameter>\n' }}
+                    {%- endfor %}
+                {%- endif %}
+                {{- '</function>\n</tool_call>' }}
+            {%- endfor %}
+        {%- endif %}
+        {{- '<|im_end|>\n' }}
+    {%- elif message.role == "tool" %}
+        {%- if loop.previtem and loop.previtem.role != "tool" %}
+            {{- '<|im_start|>user' }}
+        {%- endif %}
+        {{- '\n<tool_response>\n' }}
+        {{- content }}
+        {{- '\n</tool_response>' }}
+        {%- if not loop.last and loop.nextitem.role != "tool" %}
+            {{- '<|im_end|>\n' }}
+        {%- elif loop.last %}
+            {{- '<|im_end|>\n' }}
+        {%- endif %}
+    {%- else %}
+        {{- raise_exception('Unexpected message role.') }}
+    {%- endif %}
+{%- endfor %}
+{%- if add_generation_prompt %}
+    {{- '<|im_start|>assistant\n' }}
+    {%- if enable_thinking is defined and enable_thinking is false %}
+        {{- '<think>\n\n</think>\n\n' }}
+    {%- else %}
+        {{- '<think>\n' }}
+    {%- endif %}
+{%- endif %}

overgeneralisation_original_Estonian/Qwen3.5-4B-Base_overgeneralisation_splits_original_features_train_overgeneralisation_splits_original_features_test2/checkpoint-1080/tokenizer_config.json

@@ -0,0 +1,31 @@
+{
+  "add_prefix_space": false,
+  "audio_bos_token": "<|audio_start|>",
+  "audio_eos_token": "<|audio_end|>",
+  "audio_token": "<|audio_pad|>",
+  "backend": "tokenizers",
+  "bos_token": null,
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<|endoftext|>",
+  "errors": "replace",
+  "image_token": "<|image_pad|>",
+  "is_local": false,
+  "model_max_length": 262144,
+  "model_specific_special_tokens": {
+    "audio_bos_token": "<|audio_start|>",
+    "audio_eos_token": "<|audio_end|>",
+    "audio_token": "<|audio_pad|>",
+    "image_token": "<|image_pad|>",
+    "video_token": "<|video_pad|>",
+    "vision_bos_token": "<|vision_start|>",
+    "vision_eos_token": "<|vision_end|>"
+  },
+  "pad_token": "<|endoftext|>",
+  "pretokenize_regex": "(?i:'s|'t|'re|'ve|'m|'ll|'d)|[^\\r\\n\\p{L}\\p{N}]?[\\p{L}\\p{M}]+|\\p{N}| ?[^\\s\\p{L}\\p{M}\\p{N}]+[\\r\\n]*|\\s*[\\r\\n]+|\\s+(?!\\S)|\\s+",
+  "split_special_tokens": false,
+  "tokenizer_class": "TokenizersBackend",
+  "unk_token": null,
+  "video_token": "<|video_pad|>",
+  "vision_bos_token": "<|vision_start|>",
+  "vision_eos_token": "<|vision_end|>"
+}

overgeneralisation_original_Estonian/Qwen3.5-4B-Base_overgeneralisation_splits_original_features_train_overgeneralisation_splits_original_features_test2/checkpoint-1080/trainer_state.json

@@ -0,0 +1,1168 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 2.603500301750151,
+  "eval_steps": 20,
+  "global_step": 1080,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "entropy": 2.172999547421932,
+      "epoch": 0.04828002414001207,
+      "grad_norm": 2.5071208477020264,
+      "learning_rate": 1.7227585487708594e-05,
+      "loss": 1.9347333908081055,
+      "mean_token_accuracy": 0.6086816020309925,
+      "num_tokens": 47778.0,
+      "step": 20
+    },
+    {
+      "epoch": 0.04828002414001207,
+      "eval_entropy": 1.6909754791956269,
+      "eval_loss": 1.4728385210037231,
+      "eval_mean_token_accuracy": 0.6773593797442619,
+      "eval_num_tokens": 47778.0,
+      "eval_runtime": 90.1265,
+      "eval_samples_per_second": 15.756,
+      "eval_steps_per_second": 1.975,
+      "step": 20
+    },
+    {
+      "entropy": 1.2014626950025558,
+      "epoch": 0.09656004828002414,
+      "grad_norm": 1.5074543952941895,
+      "learning_rate": 3.5361886001086065e-05,
+      "loss": 1.0867146492004394,
+      "mean_token_accuracy": 0.7285927519202232,
+      "num_tokens": 95981.0,
+      "step": 40
+    },
+    {
+      "epoch": 0.09656004828002414,
+      "eval_entropy": 1.0304168352250302,
+      "eval_loss": 0.9211422204971313,
+      "eval_mean_token_accuracy": 0.7613122848312507,
+      "eval_num_tokens": 95981.0,
+      "eval_runtime": 88.6338,
+      "eval_samples_per_second": 16.021,
+      "eval_steps_per_second": 2.008,
+      "step": 40
+    },
+    {
+      "entropy": 0.9454198732972146,
+      "epoch": 0.14484007242003621,
+      "grad_norm": 1.7578108310699463,
+      "learning_rate": 5.349618651446354e-05,
+      "loss": 0.8447297096252442,
+      "mean_token_accuracy": 0.7723424732685089,
+      "num_tokens": 142784.0,
+      "step": 60
+    },
+    {
+      "epoch": 0.14484007242003621,
+      "eval_entropy": 0.8647834474451086,
+      "eval_loss": 0.8240737318992615,
+      "eval_mean_token_accuracy": 0.7821227014064789,
+      "eval_num_tokens": 142784.0,
+      "eval_runtime": 88.0135,
+      "eval_samples_per_second": 16.134,
+      "eval_steps_per_second": 2.022,
+      "step": 60
+    },
+    {
+      "entropy": 0.8963550612330436,
+      "epoch": 0.19312009656004828,
+      "grad_norm": 1.1464931964874268,
+      "learning_rate": 7.1630487027841e-05,
+      "loss": 0.7979836463928223,
+      "mean_token_accuracy": 0.7822489373385906,
+      "num_tokens": 187338.0,
+      "step": 80
+    },
+    {
+      "epoch": 0.19312009656004828,
+      "eval_entropy": 0.8663296133614657,
+      "eval_loss": 0.7752988338470459,
+      "eval_mean_token_accuracy": 0.7885957012015782,
+      "eval_num_tokens": 187338.0,
+      "eval_runtime": 88.284,
+      "eval_samples_per_second": 16.084,
+      "eval_steps_per_second": 2.016,
+      "step": 80
+    },
+    {
+      "entropy": 0.8265103206038475,
+      "epoch": 0.24140012070006034,
+      "grad_norm": 1.0106662511825562,
+      "learning_rate": 8.976478754121848e-05,
+      "loss": 0.7297060012817382,
+      "mean_token_accuracy": 0.7955138415098191,
+      "num_tokens": 236038.0,
+      "step": 100
+    },
+    {
+      "epoch": 0.24140012070006034,
+      "eval_entropy": 0.8122169536151244,
+      "eval_loss": 0.754906177520752,
+      "eval_mean_token_accuracy": 0.7917155255092664,
+      "eval_num_tokens": 236038.0,
+      "eval_runtime": 87.5754,
+      "eval_samples_per_second": 16.215,
+      "eval_steps_per_second": 2.033,
+      "step": 100
+    },
+    {
+      "entropy": 0.8098392844200134,
+      "epoch": 0.28968014484007243,
+      "grad_norm": 1.2238041162490845,
+      "learning_rate": 0.00010789908805459595,
+      "loss": 0.7201489448547364,
+      "mean_token_accuracy": 0.7975016921758652,
+      "num_tokens": 282918.0,
+      "step": 120
+    },
+    {
+      "epoch": 0.28968014484007243,
+      "eval_entropy": 0.8418726462326692,
+      "eval_loss": 0.7393178939819336,
+      "eval_mean_token_accuracy": 0.7936771585700217,
+      "eval_num_tokens": 282918.0,
+      "eval_runtime": 87.1597,
+      "eval_samples_per_second": 16.292,
+      "eval_steps_per_second": 2.042,
+      "step": 120
+    },
+    {
+      "entropy": 0.8080591082572937,
+      "epoch": 0.33796016898008446,
+      "grad_norm": 1.011591911315918,
+      "learning_rate": 0.00012603338856797343,
+      "loss": 0.7147697925567627,
+      "mean_token_accuracy": 0.7947175532579422,
+      "num_tokens": 331305.0,
+      "step": 140
+    },
+    {
+      "epoch": 0.33796016898008446,
+      "eval_entropy": 0.7894941431083037,
+      "eval_loss": 0.720219075679779,
+      "eval_mean_token_accuracy": 0.801899236239744,
+      "eval_num_tokens": 331305.0,
+      "eval_runtime": 87.1493,
+      "eval_samples_per_second": 16.294,
+      "eval_steps_per_second": 2.042,
+      "step": 140
+    },
+    {
+      "entropy": 0.7957205109298229,
+      "epoch": 0.38624019312009655,
+      "grad_norm": 0.8481830954551697,
+      "learning_rate": 0.00014416768908135088,
+      "loss": 0.7050027847290039,
+      "mean_token_accuracy": 0.7995368830859662,
+      "num_tokens": 375876.0,
+      "step": 160
+    },
+    {
+      "epoch": 0.38624019312009655,
+      "eval_entropy": 0.7746140591883928,
+      "eval_loss": 0.7175475358963013,
+      "eval_mean_token_accuracy": 0.7994848955213354,
+      "eval_num_tokens": 375876.0,
+      "eval_runtime": 86.655,
+      "eval_samples_per_second": 16.387,
+      "eval_steps_per_second": 2.054,
+      "step": 160
+    },
+    {
+      "entropy": 0.779565304517746,
+      "epoch": 0.43452021726010864,
+      "grad_norm": 1.032624363899231,
+      "learning_rate": 0.00016230198959472835,
+      "loss": 0.6978332042694092,
+      "mean_token_accuracy": 0.8015618294477462,
+      "num_tokens": 423913.0,
+      "step": 180
+    },
+    {
+      "epoch": 0.43452021726010864,
+      "eval_entropy": 0.7455583871080634,
+      "eval_loss": 0.7118850350379944,
+      "eval_mean_token_accuracy": 0.7999682195400923,
+      "eval_num_tokens": 423913.0,
+      "eval_runtime": 86.7842,
+      "eval_samples_per_second": 16.362,
+      "eval_steps_per_second": 2.051,
+      "step": 180
+    },
+    {
+      "entropy": 0.7810172818601131,
+      "epoch": 0.4828002414001207,
+      "grad_norm": 0.9276663064956665,
+      "learning_rate": 0.00018043629010810582,
+      "loss": 0.6937861442565918,
+      "mean_token_accuracy": 0.802893053740263,
+      "num_tokens": 472395.0,
+      "step": 200
+    },
+    {
+      "epoch": 0.4828002414001207,
+      "eval_entropy": 0.7650193290764027,
+      "eval_loss": 0.7020567059516907,
+      "eval_mean_token_accuracy": 0.8045302153973097,
+      "eval_num_tokens": 472395.0,
+      "eval_runtime": 88.326,
+      "eval_samples_per_second": 16.077,
+      "eval_steps_per_second": 2.015,
+      "step": 200
+    },
+    {
+      "entropy": 0.7862060204148292,
+      "epoch": 0.5310802655401328,
+      "grad_norm": 0.8195295929908752,
+      "learning_rate": 0.0001985705906214833,
+      "loss": 0.6878782272338867,
+      "mean_token_accuracy": 0.8027168907225132,
+      "num_tokens": 517049.0,
+      "step": 220
+    },
+    {
+      "epoch": 0.5310802655401328,
+      "eval_entropy": 0.7656565583154057,
+      "eval_loss": 0.6973932981491089,
+      "eval_mean_token_accuracy": 0.8047967278555538,
+      "eval_num_tokens": 517049.0,
+      "eval_runtime": 87.7698,
+      "eval_samples_per_second": 16.179,
+      "eval_steps_per_second": 2.028,
+      "step": 220
+    },
+    {
+      "entropy": 0.7611534461379051,
+      "epoch": 0.5793602896801449,
+      "grad_norm": 0.7499712109565735,
+      "learning_rate": 0.00021670489113486077,
+      "loss": 0.681937837600708,
+      "mean_token_accuracy": 0.8041978515684605,
+      "num_tokens": 562836.0,
+      "step": 240
+    },
+    {
+      "epoch": 0.5793602896801449,
+      "eval_entropy": 0.7599442380197933,
+      "eval_loss": 0.7002046704292297,
+      "eval_mean_token_accuracy": 0.8051861517884759,
+      "eval_num_tokens": 562836.0,
+      "eval_runtime": 87.9239,
+      "eval_samples_per_second": 16.15,
+      "eval_steps_per_second": 2.024,
+      "step": 240
+    },
+    {
+      "entropy": 0.7800783462822437,
+      "epoch": 0.627640313820157,
+      "grad_norm": 0.8645269274711609,
+      "learning_rate": 0.00023483919164823825,
+      "loss": 0.6954500675201416,
+      "mean_token_accuracy": 0.8031484372913837,
+      "num_tokens": 606612.0,
+      "step": 260
+    },
+    {
+      "epoch": 0.627640313820157,
+      "eval_entropy": 0.7498679090751691,
+      "eval_loss": 0.7046149373054504,
+      "eval_mean_token_accuracy": 0.802388215667746,
+      "eval_num_tokens": 606612.0,
+      "eval_runtime": 88.2468,
+      "eval_samples_per_second": 16.091,
+      "eval_steps_per_second": 2.017,
+      "step": 260
+    },
+    {
+      "entropy": 0.7586533665657044,
+      "epoch": 0.6759203379601689,
+      "grad_norm": 0.741596519947052,
+      "learning_rate": 0.0002529734921616157,
+      "loss": 0.690476369857788,
+      "mean_token_accuracy": 0.8046923100948333,
+      "num_tokens": 655920.0,
+      "step": 280
+    },
+    {
+      "epoch": 0.6759203379601689,
+      "eval_entropy": 0.7714274066887544,
+      "eval_loss": 0.6992160677909851,
+      "eval_mean_token_accuracy": 0.8042646482419432,
+      "eval_num_tokens": 655920.0,
+      "eval_runtime": 88.1754,
+      "eval_samples_per_second": 16.104,
+      "eval_steps_per_second": 2.019,
+      "step": 280
+    },
+    {
+      "entropy": 0.7585848286747933,
+      "epoch": 0.724200362100181,
+      "grad_norm": 1.0778124332427979,
+      "learning_rate": 0.00027110779267499317,
+      "loss": 0.6790409564971924,
+      "mean_token_accuracy": 0.8060351841151714,
+      "num_tokens": 702638.0,
+      "step": 300
+    },
+    {
+      "epoch": 0.724200362100181,
+      "eval_entropy": 0.7380402811457601,
+      "eval_loss": 0.7054820656776428,
+      "eval_mean_token_accuracy": 0.8003534738267406,
+      "eval_num_tokens": 702638.0,
+      "eval_runtime": 88.238,
+      "eval_samples_per_second": 16.093,
+      "eval_steps_per_second": 2.017,
+      "step": 300
+    },
+    {
+      "entropy": 0.7696981988847256,
+      "epoch": 0.7724803862401931,
+      "grad_norm": 0.8408658504486084,
+      "learning_rate": 0.00028924209318837064,
+      "loss": 0.7070387363433838,
+      "mean_token_accuracy": 0.8035947173833847,
+      "num_tokens": 749377.0,
+      "step": 320
+    },
+    {
+      "epoch": 0.7724803862401931,
+      "eval_entropy": 0.7897409023193831,
+      "eval_loss": 0.7015668749809265,
+      "eval_mean_token_accuracy": 0.805306687113944,
+      "eval_num_tokens": 749377.0,
+      "eval_runtime": 87.9395,
+      "eval_samples_per_second": 16.147,
+      "eval_steps_per_second": 2.024,
+      "step": 320
+    },
+    {
+      "entropy": 0.762337189912796,
+      "epoch": 0.8207604103802052,
+      "grad_norm": 0.9843310117721558,
+      "learning_rate": 0.0003073763937017481,
+      "loss": 0.7021088123321533,
+      "mean_token_accuracy": 0.8029616877436638,
+      "num_tokens": 798874.0,
+      "step": 340
+    },
+    {
+      "epoch": 0.8207604103802052,
+      "eval_entropy": 0.7210077121016685,
+      "eval_loss": 0.7082746624946594,
+      "eval_mean_token_accuracy": 0.8049283824609906,
+      "eval_num_tokens": 798874.0,
+      "eval_runtime": 88.1706,
+      "eval_samples_per_second": 16.105,
+      "eval_steps_per_second": 2.019,
+      "step": 340
+    },
+    {
+      "entropy": 0.7879025422036647,
+      "epoch": 0.8690404345202173,
+      "grad_norm": 0.9713814854621887,
+      "learning_rate": 0.0003255106942151256,
+      "loss": 0.714624547958374,
+      "mean_token_accuracy": 0.79987031519413,
+      "num_tokens": 840825.0,
+      "step": 360
+    },
+    {
+      "epoch": 0.8690404345202173,
+      "eval_entropy": 0.7866930489459735,
+      "eval_loss": 0.7145519852638245,
+      "eval_mean_token_accuracy": 0.8015657988157165,
+      "eval_num_tokens": 840825.0,
+      "eval_runtime": 88.3746,
+      "eval_samples_per_second": 16.068,
+      "eval_steps_per_second": 2.014,
+      "step": 360
+    },
+    {
+      "entropy": 0.7844726584851742,
+      "epoch": 0.9173204586602294,
+      "grad_norm": 1.2028311491012573,
+      "learning_rate": 0.00034364499472850306,
+      "loss": 0.7040786266326904,
+      "mean_token_accuracy": 0.7981885217130185,
+      "num_tokens": 882885.0,
+      "step": 380
+    },
+    {
+      "epoch": 0.9173204586602294,
+      "eval_entropy": 0.7767668339643585,
+      "eval_loss": 0.7073574066162109,
+      "eval_mean_token_accuracy": 0.8040851833445303,
+      "eval_num_tokens": 882885.0,
+      "eval_runtime": 88.3828,
+      "eval_samples_per_second": 16.066,
+      "eval_steps_per_second": 2.014,
+      "step": 380
+    },
+    {
+      "entropy": 0.7699790760874748,
+      "epoch": 0.9656004828002414,
+      "grad_norm": 0.9397398233413696,
+      "learning_rate": 0.00036177929524188054,
+      "loss": 0.7040313720703125,
+      "mean_token_accuracy": 0.8034727744758129,
+      "num_tokens": 927456.0,
+      "step": 400
+    },
+    {
+      "epoch": 0.9656004828002414,
+      "eval_entropy": 0.7749103545472863,
+      "eval_loss": 0.7093513011932373,
+      "eval_mean_token_accuracy": 0.8019482901926791,
+      "eval_num_tokens": 927456.0,
+      "eval_runtime": 86.2112,
+      "eval_samples_per_second": 16.471,
+      "eval_steps_per_second": 2.065,
+      "step": 400
+    },
+    {
+      "entropy": 0.7672389172888422,
+      "epoch": 1.012070006035003,
+      "grad_norm": 0.99210524559021,
+      "learning_rate": 0.00037628567078152296,
+      "loss": 0.6913118362426758,
+      "mean_token_accuracy": 0.8042210944287189,
+      "num_tokens": 972662.0,
+      "step": 420
+    },
+    {
+      "epoch": 1.012070006035003,
+      "eval_entropy": 0.7294771229283193,
+      "eval_loss": 0.7171520590782166,
+      "eval_mean_token_accuracy": 0.8015128585059991,
+      "eval_num_tokens": 972662.0,
+      "eval_runtime": 88.037,
+      "eval_samples_per_second": 16.13,
+      "eval_steps_per_second": 2.022,
+      "step": 420
+    },
+    {
+      "entropy": 0.7143943183124065,
+      "epoch": 1.060350030175015,
+      "grad_norm": 1.087638258934021,
+      "learning_rate": 0.0003762484015601069,
+      "loss": 0.642704439163208,
+      "mean_token_accuracy": 0.8124966286122799,
+      "num_tokens": 1022525.0,
+      "step": 440
+    },
+    {
+      "epoch": 1.060350030175015,
+      "eval_entropy": 0.743511886744017,
+      "eval_loss": 0.7116662859916687,
+      "eval_mean_token_accuracy": 0.8037210672758939,
+      "eval_num_tokens": 1022525.0,
+      "eval_runtime": 87.9931,
+      "eval_samples_per_second": 16.138,
+      "eval_steps_per_second": 2.023,
+      "step": 440
+    },
+    {
+      "entropy": 0.7237346574664116,
+      "epoch": 1.1086300543150271,
+      "grad_norm": 0.9365553259849548,
+      "learning_rate": 0.0003761579008392123,
+      "loss": 0.652644681930542,
+      "mean_token_accuracy": 0.8107496216893196,
+      "num_tokens": 1069194.0,
+      "step": 460
+    },
+    {
+      "epoch": 1.1086300543150271,
+      "eval_entropy": 0.7372823839776972,
+      "eval_loss": 0.7238303422927856,
+      "eval_mean_token_accuracy": 0.8026230124275336,
+      "eval_num_tokens": 1069194.0,
+      "eval_runtime": 87.6778,
+      "eval_samples_per_second": 16.196,
+      "eval_steps_per_second": 2.03,
+      "step": 460
+    },
+    {
+      "entropy": 0.7202137842774391,
+      "epoch": 1.1569100784550392,
+      "grad_norm": 0.7998443841934204,
+      "learning_rate": 0.0003760141942294726,
+      "loss": 0.6485219478607178,
+      "mean_token_accuracy": 0.8092714451253414,
+      "num_tokens": 1117570.0,
+      "step": 480
+    },
+    {
+      "epoch": 1.1569100784550392,
+      "eval_entropy": 0.71547647942318,
+      "eval_loss": 0.7124277949333191,
+      "eval_mean_token_accuracy": 0.8013174219077892,
+      "eval_num_tokens": 1117570.0,
+      "eval_runtime": 87.4387,
+      "eval_samples_per_second": 16.24,
+      "eval_steps_per_second": 2.036,
+      "step": 480
+    },
+    {
+      "entropy": 0.7270819500088692,
+      "epoch": 1.2051901025950513,
+      "grad_norm": 0.9470787644386292,
+      "learning_rate": 0.00037581732239815854,
+      "loss": 0.66352219581604,
+      "mean_token_accuracy": 0.8102593503892421,
+      "num_tokens": 1161717.0,
+      "step": 500
+    },
+    {
+      "epoch": 1.2051901025950513,
+      "eval_entropy": 0.7791565897759427,
+      "eval_loss": 0.7179726958274841,
+      "eval_mean_token_accuracy": 0.7948898328154275,
+      "eval_num_tokens": 1161717.0,
+      "eval_runtime": 87.9927,
+      "eval_samples_per_second": 16.138,
+      "eval_steps_per_second": 2.023,
+      "step": 500
+    },
+    {
+      "entropy": 0.7320936664938926,
+      "epoch": 1.2534701267350634,
+      "grad_norm": 1.0220379829406738,
+      "learning_rate": 0.0003755673410576695,
+      "loss": 0.6581549644470215,
+      "mean_token_accuracy": 0.8098709337413311,
+      "num_tokens": 1209390.0,
+      "step": 520
+    },
+    {
+      "epoch": 1.2534701267350634,
+      "eval_entropy": 0.7710678225822662,
+      "eval_loss": 0.708797812461853,
+      "eval_mean_token_accuracy": 0.8039572778042783,
+      "eval_num_tokens": 1209390.0,
+      "eval_runtime": 88.1193,
+      "eval_samples_per_second": 16.115,
+      "eval_steps_per_second": 2.02,
+      "step": 520
+    },
+    {
+      "entropy": 0.7281674664467573,
+      "epoch": 1.3017501508750755,
+      "grad_norm": 0.8775188326835632,
+      "learning_rate": 0.000375264320949768,
+      "loss": 0.6631278991699219,
+      "mean_token_accuracy": 0.8101261422038079,
+      "num_tokens": 1256611.0,
+      "step": 540
+    },
+    {
+      "epoch": 1.3017501508750755,
+      "eval_entropy": 0.7199799800187014,
+      "eval_loss": 0.6981531381607056,
+      "eval_mean_token_accuracy": 0.8071265254127845,
+      "eval_num_tokens": 1256611.0,
+      "eval_runtime": 88.165,
+      "eval_samples_per_second": 16.106,
+      "eval_steps_per_second": 2.019,
+      "step": 540
+    },
+    {
+      "entropy": 0.722860773652792,
+      "epoch": 1.3500301750150876,
+      "grad_norm": 0.8949288129806519,
+      "learning_rate": 0.00037490834782556,
+      "loss": 0.6639598846435547,
+      "mean_token_accuracy": 0.8105649061501026,
+      "num_tokens": 1302325.0,
+      "step": 560
+    },
+    {
+      "epoch": 1.3500301750150876,
+      "eval_entropy": 0.7134760032209118,
+      "eval_loss": 0.7029948234558105,
+      "eval_mean_token_accuracy": 0.8062985880991046,
+      "eval_num_tokens": 1302325.0,
+      "eval_runtime": 88.1286,
+      "eval_samples_per_second": 16.113,
+      "eval_steps_per_second": 2.02,
+      "step": 560
+    },
+    {
+      "entropy": 0.7299126699566841,
+      "epoch": 1.3983101991550995,
+      "grad_norm": 1.0323039293289185,
+      "learning_rate": 0.0003744995224212291,
+      "loss": 0.6535763263702392,
+      "mean_token_accuracy": 0.8120224848389626,
+      "num_tokens": 1343844.0,
+      "step": 580
+    },
+    {
+      "epoch": 1.3983101991550995,
+      "eval_entropy": 0.7169701182440426,
+      "eval_loss": 0.7067857980728149,
+      "eval_mean_token_accuracy": 0.8040432936689826,
+      "eval_num_tokens": 1343844.0,
+      "eval_runtime": 88.2252,
+      "eval_samples_per_second": 16.095,
+      "eval_steps_per_second": 2.018,
+      "step": 580
+    },
+    {
+      "entropy": 0.7281742256134749,
+      "epoch": 1.4465902232951118,
+      "grad_norm": 1.3370424509048462,
+      "learning_rate": 0.00037403796042952863,
+      "loss": 0.6637272357940673,
+      "mean_token_accuracy": 0.8118261776864528,
+      "num_tokens": 1389471.0,
+      "step": 600
+    },
+    {
+      "epoch": 1.4465902232951118,
+      "eval_entropy": 0.7378843212395572,
+      "eval_loss": 0.7038390040397644,
+      "eval_mean_token_accuracy": 0.8059229371922739,
+      "eval_num_tokens": 1389471.0,
+      "eval_runtime": 87.9301,
+      "eval_samples_per_second": 16.149,
+      "eval_steps_per_second": 2.024,
+      "step": 600
+    },
+    {
+      "entropy": 0.7388614989817143,
+      "epoch": 1.4948702474351236,
+      "grad_norm": 0.8464171290397644,
+      "learning_rate": 0.00037352379246704257,
+      "loss": 0.6604740142822265,
+      "mean_token_accuracy": 0.8102974124252796,
+      "num_tokens": 1435975.0,
+      "step": 620
+    },
+    {
+      "epoch": 1.4948702474351236,
+      "eval_entropy": 0.7126847046814607,
+      "eval_loss": 0.6950703263282776,
+      "eval_mean_token_accuracy": 0.8085885392815879,
+      "eval_num_tokens": 1435975.0,
+      "eval_runtime": 88.2732,
+      "eval_samples_per_second": 16.086,
+      "eval_steps_per_second": 2.016,
+      "step": 620
+    },
+    {
+      "entropy": 0.7258916139602661,
+      "epoch": 1.5431502715751357,
+      "grad_norm": 1.1780085563659668,
+      "learning_rate": 0.0003729571640372223,
+      "loss": 0.6529891014099121,
+      "mean_token_accuracy": 0.8085981778800487,
+      "num_tokens": 1482035.0,
+      "step": 640
+    },
+    {
+      "epoch": 1.5431502715751357,
+      "eval_entropy": 0.7096509638797032,
+      "eval_loss": 0.690051257610321,
+      "eval_mean_token_accuracy": 0.8100520241796301,
+      "eval_num_tokens": 1482035.0,
+      "eval_runtime": 88.3153,
+      "eval_samples_per_second": 16.079,
+      "eval_steps_per_second": 2.016,
+      "step": 640
+    },
+    {
+      "entropy": 0.7151173129677773,
+      "epoch": 1.5914302957151478,
+      "grad_norm": 1.0357667207717896,
+      "learning_rate": 0.0003723382354892108,
+      "loss": 0.6451474189758301,
+      "mean_token_accuracy": 0.8136902332305909,
+      "num_tokens": 1528119.0,
+      "step": 660
+    },
+    {
+      "epoch": 1.5914302957151478,
+      "eval_entropy": 0.7060252652409371,
+      "eval_loss": 0.6899483799934387,
+      "eval_mean_token_accuracy": 0.810834194502134,
+      "eval_num_tokens": 1528119.0,
+      "eval_runtime": 87.8848,
+      "eval_samples_per_second": 16.158,
+      "eval_steps_per_second": 2.025,
+      "step": 660
+    },
+    {
+      "entropy": 0.7141377851366997,
+      "epoch": 1.63971031985516,
+      "grad_norm": 0.8391667008399963,
+      "learning_rate": 0.000371667181972466,
+      "loss": 0.6481579780578614,
+      "mean_token_accuracy": 0.8159015104174614,
+      "num_tokens": 1573661.0,
+      "step": 680
+    },
+    {
+      "epoch": 1.63971031985516,
+      "eval_entropy": 0.7598993319473909,
+      "eval_loss": 0.681204080581665,
+      "eval_mean_token_accuracy": 0.8100194586126992,
+      "eval_num_tokens": 1573661.0,
+      "eval_runtime": 88.2415,
+      "eval_samples_per_second": 16.092,
+      "eval_steps_per_second": 2.017,
+      "step": 680
+    },
+    {
+      "entropy": 0.7235980287194252,
+      "epoch": 1.687990343995172,
+      "grad_norm": 0.8197196125984192,
+      "learning_rate": 0.00037094419338719537,
+      "loss": 0.6582849025726318,
+      "mean_token_accuracy": 0.8114834539592266,
+      "num_tokens": 1621714.0,
+      "step": 700
+    },
+    {
+      "epoch": 1.687990343995172,
+      "eval_entropy": 0.7166854568411795,
+      "eval_loss": 0.679177463054657,
+      "eval_mean_token_accuracy": 0.8119039877077167,
+      "eval_num_tokens": 1621714.0,
+      "eval_runtime": 88.0957,
+      "eval_samples_per_second": 16.119,
+      "eval_steps_per_second": 2.021,
+      "step": 700
+    },
+    {
+      "entropy": 0.7278190158307553,
+      "epoch": 1.736270368135184,
+      "grad_norm": 0.7820602655410767,
+      "learning_rate": 0.0003701694743306164,
+      "loss": 0.6633153438568116,
+      "mean_token_accuracy": 0.8123249113559723,
+      "num_tokens": 1672118.0,
+      "step": 720
+    },
+    {
+      "epoch": 1.736270368135184,
+      "eval_entropy": 0.6980850314826108,
+      "eval_loss": 0.6775653958320618,
+      "eval_mean_token_accuracy": 0.8116872906684875,
+      "eval_num_tokens": 1672118.0,
+      "eval_runtime": 87.9795,
+      "eval_samples_per_second": 16.14,
+      "eval_steps_per_second": 2.023,
+      "step": 720
+    },
+    {
+      "entropy": 0.7149631556123495,
+      "epoch": 1.7845503922751962,
+      "grad_norm": 1.0139760971069336,
+      "learning_rate": 0.00036934324403905795,
+      "loss": 0.6468976020812989,
+      "mean_token_accuracy": 0.8117561548948288,
+      "num_tokens": 1720465.0,
+      "step": 740
+    },
+    {
+      "epoch": 1.7845503922751962,
+      "eval_entropy": 0.69009849094273,
+      "eval_loss": 0.6731093525886536,
+      "eval_mean_token_accuracy": 0.81281964196248,
+      "eval_num_tokens": 1720465.0,
+      "eval_runtime": 87.7412,
+      "eval_samples_per_second": 16.184,
+      "eval_steps_per_second": 2.029,
+      "step": 740
+    },
+    {
+      "entropy": 0.7226049453020096,
+      "epoch": 1.832830416415208,
+      "grad_norm": 1.5040546655654907,
+      "learning_rate": 0.0003684657363259193,
+      "loss": 0.6385328769683838,
+      "mean_token_accuracy": 0.8132639616727829,
+      "num_tokens": 1764440.0,
+      "step": 760
+    },
+    {
+      "epoch": 1.832830416415208,
+      "eval_entropy": 0.7125169618076153,
+      "eval_loss": 0.6864680647850037,
+      "eval_mean_token_accuracy": 0.8049418089095126,
+      "eval_num_tokens": 1764440.0,
+      "eval_runtime": 87.9429,
+      "eval_samples_per_second": 16.147,
+      "eval_steps_per_second": 2.024,
+      "step": 760
+    },
+    {
+      "entropy": 0.707821810990572,
+      "epoch": 1.8811104405552204,
+      "grad_norm": 0.8604223132133484,
+      "learning_rate": 0.00036753719951550327,
+      "loss": 0.6397049427032471,
+      "mean_token_accuracy": 0.8137877315282822,
+      "num_tokens": 1810828.0,
+      "step": 780
+    },
+    {
+      "epoch": 1.8811104405552204,
+      "eval_entropy": 0.6492102551326323,
+      "eval_loss": 0.6743726134300232,
+      "eval_mean_token_accuracy": 0.8130356572317273,
+      "eval_num_tokens": 1810828.0,
+      "eval_runtime": 87.8684,
+      "eval_samples_per_second": 16.161,
+      "eval_steps_per_second": 2.026,
+      "step": 780
+    },
+    {
+      "entropy": 0.6800355084240437,
+      "epoch": 1.9293904646952322,
+      "grad_norm": 0.7270982265472412,
+      "learning_rate": 0.00036655789637274377,
+      "loss": 0.6267755031585693,
+      "mean_token_accuracy": 0.8180945307016373,
+      "num_tokens": 1857600.0,
+      "step": 800
+    },
+    {
+      "epoch": 1.9293904646952322,
+      "eval_entropy": 0.6727538651294922,
+      "eval_loss": 0.6696028709411621,
+      "eval_mean_token_accuracy": 0.8116443090224534,
+      "eval_num_tokens": 1857600.0,
+      "eval_runtime": 88.2449,
+      "eval_samples_per_second": 16.092,
+      "eval_steps_per_second": 2.017,
+      "step": 800
+    },
+    {
+      "entropy": 0.6746691606938839,
+      "epoch": 1.9776704888352445,
+      "grad_norm": 0.9771540760993958,
+      "learning_rate": 0.0003655281040288461,
+      "loss": 0.6326992988586426,
+      "mean_token_accuracy": 0.8179976396262646,
+      "num_tokens": 1901998.0,
+      "step": 820
+    },
+    {
+      "epoch": 1.9776704888352445,
+      "eval_entropy": 0.6636319716324967,
+      "eval_loss": 0.6714363098144531,
+      "eval_mean_token_accuracy": 0.8126635943235976,
+      "eval_num_tokens": 1901998.0,
+      "eval_runtime": 87.7666,
+      "eval_samples_per_second": 16.179,
+      "eval_steps_per_second": 2.028,
+      "step": 820
+    },
+    {
+      "entropy": 0.6319125778489298,
+      "epoch": 2.024140012070006,
+      "grad_norm": 0.8380444645881653,
+      "learning_rate": 0.0003644481139028622,
+      "loss": 0.5780903816223144,
+      "mean_token_accuracy": 0.8250188401767186,
+      "num_tokens": 1947553.0,
+      "step": 840
+    },
+    {
+      "epoch": 2.024140012070006,
+      "eval_entropy": 0.6627151989535,
+      "eval_loss": 0.6774935722351074,
+      "eval_mean_token_accuracy": 0.8140510819601209,
+      "eval_num_tokens": 1947553.0,
+      "eval_runtime": 87.8287,
+      "eval_samples_per_second": 16.168,
+      "eval_steps_per_second": 2.027,
+      "step": 840
+    },
+    {
+      "entropy": 0.6059147048741579,
+      "epoch": 2.0724200362100182,
+      "grad_norm": 0.9745152592658997,
+      "learning_rate": 0.0003633182316192229,
+      "loss": 0.546702527999878,
+      "mean_token_accuracy": 0.8358186542987823,
+      "num_tokens": 1995234.0,
+      "step": 860
+    },
+    {
+      "epoch": 2.0724200362100182,
+      "eval_entropy": 0.6620710384979677,
+      "eval_loss": 0.6769291162490845,
+      "eval_mean_token_accuracy": 0.8135082835561773,
+      "eval_num_tokens": 1995234.0,
+      "eval_runtime": 88.0008,
+      "eval_samples_per_second": 16.136,
+      "eval_steps_per_second": 2.023,
+      "step": 860
+    },
+    {
+      "entropy": 0.6076117843389511,
+      "epoch": 2.12070006035003,
+      "grad_norm": 0.6914380192756653,
+      "learning_rate": 0.0003621387769212491,
+      "loss": 0.5324535846710206,
+      "mean_token_accuracy": 0.8366568259894848,
+      "num_tokens": 2040639.0,
+      "step": 880
+    },
+    {
+      "epoch": 2.12070006035003,
+      "eval_entropy": 0.6592224213514435,
+      "eval_loss": 0.670383632183075,
+      "eval_mean_token_accuracy": 0.8162150088320957,
+      "eval_num_tokens": 2040639.0,
+      "eval_runtime": 88.324,
+      "eval_samples_per_second": 16.077,
+      "eval_steps_per_second": 2.015,
+      "step": 880
+    },
+    {
+      "entropy": 0.6158072650432587,
+      "epoch": 2.1689800844900424,
+      "grad_norm": 0.9649198055267334,
+      "learning_rate": 0.0003609100835806688,
+      "loss": 0.5564568996429443,
+      "mean_token_accuracy": 0.8329654954373836,
+      "num_tokens": 2090604.0,
+      "step": 900
+    },
+    {
+      "epoch": 2.1689800844900424,
+      "eval_entropy": 0.608588819758276,
+      "eval_loss": 0.6745102405548096,
+      "eval_mean_token_accuracy": 0.8155080542135774,
+      "eval_num_tokens": 2090604.0,
+      "eval_runtime": 87.52,
+      "eval_samples_per_second": 16.225,
+      "eval_steps_per_second": 2.034,
+      "step": 900
+    },
+    {
+      "entropy": 0.5938734326511621,
+      "epoch": 2.2172601086300543,
+      "grad_norm": 0.6915557384490967,
+      "learning_rate": 0.0003596324993031632,
+      "loss": 0.5351875782012939,
+      "mean_token_accuracy": 0.8385631121695042,
+      "num_tokens": 2142246.0,
+      "step": 920
+    },
+    {
+      "epoch": 2.2172601086300543,
+      "eval_entropy": 0.6251631205001574,
+      "eval_loss": 0.6664518117904663,
+      "eval_mean_token_accuracy": 0.81711940417129,
+      "eval_num_tokens": 2142246.0,
+      "eval_runtime": 87.1792,
+      "eval_samples_per_second": 16.288,
+      "eval_steps_per_second": 2.042,
+      "step": 920
+    },
+    {
+      "entropy": 0.623173613846302,
+      "epoch": 2.2655401327700666,
+      "grad_norm": 0.8366610407829285,
+      "learning_rate": 0.00035830638562997063,
+      "loss": 0.5488213539123535,
+      "mean_token_accuracy": 0.8321483485400677,
+      "num_tokens": 2183144.0,
+      "step": 940
+    },
+    {
+      "epoch": 2.2655401327700666,
+      "eval_entropy": 0.6237085649136747,
+      "eval_loss": 0.68075031042099,
+      "eval_mean_token_accuracy": 0.814671738429016,
+      "eval_num_tokens": 2183144.0,
+      "eval_runtime": 88.0652,
+      "eval_samples_per_second": 16.124,
+      "eval_steps_per_second": 2.021,
+      "step": 940
+    },
+    {
+      "entropy": 0.6236911740154027,
+      "epoch": 2.3138201569100785,
+      "grad_norm": 0.7812452912330627,
+      "learning_rate": 0.00035693211783557416,
+      "loss": 0.5696200847625732,
+      "mean_token_accuracy": 0.82991396561265,
+      "num_tokens": 2228098.0,
+      "step": 960
+    },
+    {
+      "epoch": 2.3138201569100785,
+      "eval_entropy": 0.6327033426319615,
+      "eval_loss": 0.6742984652519226,
+      "eval_mean_token_accuracy": 0.8148086091105857,
+      "eval_num_tokens": 2228098.0,
+      "eval_runtime": 88.1711,
+      "eval_samples_per_second": 16.105,
+      "eval_steps_per_second": 2.019,
+      "step": 960
+    },
+    {
+      "entropy": 0.6218426413834095,
+      "epoch": 2.3621001810500903,
+      "grad_norm": 0.8719790577888489,
+      "learning_rate": 0.0003555100848215035,
+      "loss": 0.5582651615142822,
+      "mean_token_accuracy": 0.8323961742222309,
+      "num_tokens": 2273062.0,
+      "step": 980
+    },
+    {
+      "epoch": 2.3621001810500903,
+      "eval_entropy": 0.6338012375858393,
+      "eval_loss": 0.6685822606086731,
+      "eval_mean_token_accuracy": 0.8152180300669724,
+      "eval_num_tokens": 2273062.0,
+      "eval_runtime": 88.101,
+      "eval_samples_per_second": 16.118,
+      "eval_steps_per_second": 2.02,
+      "step": 980
+    },
+    {
+      "entropy": 0.6050681818276644,
+      "epoch": 2.4103802051901027,
+      "grad_norm": 0.9151566028594971,
+      "learning_rate": 0.00035404068900628076,
+      "loss": 0.559452486038208,
+      "mean_token_accuracy": 0.8310242861509323,
+      "num_tokens": 2319305.0,
+      "step": 1000
+    },
+    {
+      "epoch": 2.4103802051901027,
+      "eval_entropy": 0.6634890969549672,
+      "eval_loss": 0.6692465543746948,
+      "eval_mean_token_accuracy": 0.8134353673190213,
+      "eval_num_tokens": 2319305.0,
+      "eval_runtime": 87.7711,
+      "eval_samples_per_second": 16.178,
+      "eval_steps_per_second": 2.028,
+      "step": 1000
+    },
+    {
+      "entropy": 0.6169968567788601,
+      "epoch": 2.4586602293301145,
+      "grad_norm": 0.7081006765365601,
+      "learning_rate": 0.0003525243462115402,
+      "loss": 0.5589694023132324,
+      "mean_token_accuracy": 0.8357502184808254,
+      "num_tokens": 2368592.0,
+      "step": 1020
+    },
+    {
+      "epoch": 2.4586602293301145,
+      "eval_entropy": 0.6126393578695447,
+      "eval_loss": 0.6706712245941162,
+      "eval_mean_token_accuracy": 0.8155577500884452,
+      "eval_num_tokens": 2368592.0,
+      "eval_runtime": 88.1407,
+      "eval_samples_per_second": 16.111,
+      "eval_steps_per_second": 2.019,
+      "step": 1020
+    },
+    {
+      "entropy": 0.6072248566895724,
+      "epoch": 2.506940253470127,
+      "grad_norm": 0.6999370455741882,
+      "learning_rate": 0.0003509614855443561,
+      "loss": 0.5500593185424805,
+      "mean_token_accuracy": 0.8337264291942119,
+      "num_tokens": 2412859.0,
+      "step": 1040
+    },
+    {
+      "epoch": 2.506940253470127,
+      "eval_entropy": 0.6417417074187418,
+      "eval_loss": 0.6651163697242737,
+      "eval_mean_token_accuracy": 0.8159113908081912,
+      "eval_num_tokens": 2412859.0,
+      "eval_runtime": 87.4854,
+      "eval_samples_per_second": 16.231,
+      "eval_steps_per_second": 2.035,
+      "step": 1040
+    },
+    {
+      "entropy": 0.6301592070609331,
+      "epoch": 2.5552202776101387,
+      "grad_norm": 0.8768910765647888,
+      "learning_rate": 0.00034935254927581064,
+      "loss": 0.5613903999328613,
+      "mean_token_accuracy": 0.8298896946012974,
+      "num_tokens": 2458201.0,
+      "step": 1060
+    },
+    {
+      "epoch": 2.5552202776101387,
+      "eval_entropy": 0.6569794751285167,
+      "eval_loss": 0.6668341159820557,
+      "eval_mean_token_accuracy": 0.8126791795987761,
+      "eval_num_tokens": 2458201.0,
+      "eval_runtime": 88.0474,
+      "eval_samples_per_second": 16.128,
+      "eval_steps_per_second": 2.022,
+      "step": 1060
+    },
+    {
+      "entropy": 0.6069021724164486,
+      "epoch": 2.603500301750151,
+      "grad_norm": 0.9033508896827698,
+      "learning_rate": 0.0003476979927158357,
+      "loss": 0.5590654373168945,
+      "mean_token_accuracy": 0.831520090252161,
+      "num_tokens": 2505382.0,
+      "step": 1080
+    },
+    {
+      "epoch": 2.603500301750151,
+      "eval_entropy": 0.6636380777600106,
+      "eval_loss": 0.6596238017082214,
+      "eval_mean_token_accuracy": 0.8177202826135614,
+      "eval_num_tokens": 2505382.0,
+      "eval_runtime": 88.08,
+      "eval_samples_per_second": 16.122,
+      "eval_steps_per_second": 2.021,
+      "step": 1080
+    }
+  ],
+  "logging_steps": 20,
+  "max_steps": 4150,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 10,
+  "save_steps": 20,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 1.0583776570652467e+17,
+  "train_batch_size": 4,
+  "trial_name": null,
+  "trial_params": null
+}

overgeneralisation_original_Estonian/Qwen3.5-4B-Base_overgeneralisation_splits_original_features_train_overgeneralisation_splits_original_features_test2/checkpoint-1100/README.md

@@ -0,0 +1,209 @@
+---
+base_model: Qwen/Qwen3.5-4B-Base
+library_name: peft
+pipeline_tag: text-generation
+tags:
+- base_model:adapter:Qwen/Qwen3.5-4B-Base
+- lora
+- sft
+- transformers
+- trl
+---
+
+# Model Card for Model ID
+
+<!-- Provide a quick summary of what the model is/does. -->
+
+
+
+## Model Details
+
+### Model Description
+
+<!-- Provide a longer summary of what this model is. -->
+
+
+
+- **Developed by:** [More Information Needed]
+- **Funded by [optional]:** [More Information Needed]
+- **Shared by [optional]:** [More Information Needed]
+- **Model type:** [More Information Needed]
+- **Language(s) (NLP):** [More Information Needed]
+- **License:** [More Information Needed]
+- **Finetuned from model [optional]:** [More Information Needed]
+
+### Model Sources [optional]
+
+<!-- Provide the basic links for the model. -->
+
+- **Repository:** [More Information Needed]
+- **Paper [optional]:** [More Information Needed]
+- **Demo [optional]:** [More Information Needed]
+
+## Uses
+
+<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
+
+### Direct Use
+
+<!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
+
+[More Information Needed]
+
+### Downstream Use [optional]
+
+<!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->
+
+[More Information Needed]
+
+### Out-of-Scope Use
+
+<!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
+
+[More Information Needed]
+
+## Bias, Risks, and Limitations
+
+<!-- This section is meant to convey both technical and sociotechnical limitations. -->
+
+[More Information Needed]
+
+### Recommendations
+
+<!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->
+
+Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
+
+## How to Get Started with the Model
+
+Use the code below to get started with the model.
+
+[More Information Needed]
+
+## Training Details
+
+### Training Data
+
+<!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
+
+[More Information Needed]
+
+### Training Procedure
+
+<!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
+
+#### Preprocessing [optional]
+
+[More Information Needed]
+
+
+#### Training Hyperparameters
+
+- **Training regime:** [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->
+
+#### Speeds, Sizes, Times [optional]
+
+<!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
+
+[More Information Needed]
+
+## Evaluation
+
+<!-- This section describes the evaluation protocols and provides the results. -->
+
+### Testing Data, Factors & Metrics
+
+#### Testing Data
+
+<!-- This should link to a Dataset Card if possible. -->
+
+[More Information Needed]
+
+#### Factors
+
+<!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->
+
+[More Information Needed]
+
+#### Metrics
+
+<!-- These are the evaluation metrics being used, ideally with a description of why. -->
+
+[More Information Needed]
+
+### Results
+
+[More Information Needed]
+
+#### Summary
+
+
+
+## Model Examination [optional]
+
+<!-- Relevant interpretability work for the model goes here -->
+
+[More Information Needed]
+
+## Environmental Impact
+
+<!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly -->
+
+Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700).
+
+- **Hardware Type:** [More Information Needed]
+- **Hours used:** [More Information Needed]
+- **Cloud Provider:** [More Information Needed]
+- **Compute Region:** [More Information Needed]
+- **Carbon Emitted:** [More Information Needed]
+
+## Technical Specifications [optional]
+
+### Model Architecture and Objective
+
+[More Information Needed]
+
+### Compute Infrastructure
+
+[More Information Needed]
+
+#### Hardware
+
+[More Information Needed]
+
+#### Software
+
+[More Information Needed]
+
+## Citation [optional]
+
+<!-- If there is a paper or blog post introducing the model, the APA and Bibtex information for that should go in this section. -->
+
+**BibTeX:**
+
+[More Information Needed]
+
+**APA:**
+
+[More Information Needed]
+
+## Glossary [optional]
+
+<!-- If relevant, include terms and calculations in this section that can help readers understand the model or model card. -->
+
+[More Information Needed]
+
+## More Information [optional]
+
+[More Information Needed]
+
+## Model Card Authors [optional]
+
+[More Information Needed]
+
+## Model Card Contact
+
+[More Information Needed]
+### Framework versions
+
+- PEFT 0.18.1

overgeneralisation_original_Estonian/Qwen3.5-4B-Base_overgeneralisation_splits_original_features_train_overgeneralisation_splits_original_features_test2/checkpoint-1100/adapter_config.json

@@ -0,0 +1,46 @@
+{
+  "alora_invocation_tokens": null,
+  "alpha_pattern": {},
+  "arrow_config": null,
+  "auto_mapping": null,
+  "base_model_name_or_path": "Qwen/Qwen3.5-4B-Base",
+  "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": 128,
+  "lora_bias": false,
+  "lora_dropout": 0.03828026524568501,
+  "megatron_config": null,
+  "megatron_core": "megatron.core",
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "peft_version": "0.18.1",
+  "qalora_group_size": 16,
+  "r": 64,
+  "rank_pattern": {},
+  "revision": null,
+  "target_modules": [
+    "gate_proj",
+    "v_proj",
+    "up_proj",
+    "down_proj",
+    "o_proj",
+    "q_proj",
+    "k_proj"
+  ],
+  "target_parameters": null,
+  "task_type": "CAUSAL_LM",
+  "trainable_token_indices": null,
+  "use_dora": false,
+  "use_qalora": false,
+  "use_rslora": false
+}

overgeneralisation_original_Estonian/Qwen3.5-4B-Base_overgeneralisation_splits_original_features_train_overgeneralisation_splits_original_features_test2/checkpoint-1100/chat_template.jinja

@@ -0,0 +1,154 @@
+{%- set image_count = namespace(value=0) %}
+{%- set video_count = namespace(value=0) %}
+{%- macro render_content(content, do_vision_count, is_system_content=false) %}
+    {%- if content is string %}
+        {{- content }}
+    {%- elif content is iterable and content is not mapping %}
+        {%- for item in content %}
+            {%- if 'image' in item or 'image_url' in item or item.type == 'image' %}
+                {%- if is_system_content %}
+                    {{- raise_exception('System message cannot contain images.') }}
+                {%- endif %}
+                {%- if do_vision_count %}
+                    {%- set image_count.value = image_count.value + 1 %}
+                {%- endif %}
+                {%- if add_vision_id %}
+                    {{- 'Picture ' ~ image_count.value ~ ': ' }}
+                {%- endif %}
+                {{- '<|vision_start|><|image_pad|><|vision_end|>' }}
+            {%- elif 'video' in item or item.type == 'video' %}
+                {%- if is_system_content %}
+                    {{- raise_exception('System message cannot contain videos.') }}
+                {%- endif %}
+                {%- if do_vision_count %}
+                    {%- set video_count.value = video_count.value + 1 %}
+                {%- endif %}
+                {%- if add_vision_id %}
+                    {{- 'Video ' ~ video_count.value ~ ': ' }}
+                {%- endif %}
+                {{- '<|vision_start|><|video_pad|><|vision_end|>' }}
+            {%- elif 'text' in item %}
+                {{- item.text }}
+            {%- else %}
+                {{- raise_exception('Unexpected item type in content.') }}
+            {%- endif %}
+        {%- endfor %}
+    {%- elif content is none or content is undefined %}
+        {{- '' }}
+    {%- else %}
+        {{- raise_exception('Unexpected content type.') }}
+    {%- endif %}
+{%- endmacro %}
+{%- if not messages %}
+    {{- raise_exception('No messages provided.') }}
+{%- endif %}
+{%- if tools and tools is iterable and tools is not mapping %}
+    {{- '<|im_start|>system\n' }}
+    {{- "# Tools\n\nYou have access to the following functions:\n\n<tools>" }}
+    {%- for tool in tools %}
+        {{- "\n" }}
+        {{- tool | tojson }}
+    {%- endfor %}
+    {{- "\n</tools>" }}
+    {{- '\n\nIf you choose to call a function ONLY reply in the following format with NO suffix:\n\n<tool_call>\n<function=example_function_name>\n<parameter=example_parameter_1>\nvalue_1\n</parameter>\n<parameter=example_parameter_2>\nThis is the value for the second parameter\nthat can span\nmultiple lines\n</parameter>\n</function>\n</tool_call>\n\n<IMPORTANT>\nReminder:\n- Function calls MUST follow the specified format: an inner <function=...></function> block must be nested within <tool_call></tool_call> XML tags\n- Required parameters MUST be specified\n- You may provide optional reasoning for your function call in natural language BEFORE the function call, but NOT after\n- If there is no function call available, answer the question like normal with your current knowledge and do not tell the user about function calls\n</IMPORTANT>' }}
+    {%- if messages[0].role == 'system' %}
+        {%- set content = render_content(messages[0].content, false, true)|trim %}
+        {%- if content %}
+            {{- '\n\n' + content }}
+        {%- endif %}
+    {%- endif %}
+    {{- '<|im_end|>\n' }}
+{%- else %}
+    {%- if messages[0].role == 'system' %}
+        {%- set content = render_content(messages[0].content, false, true)|trim %}
+        {{- '<|im_start|>system\n' + content + '<|im_end|>\n' }}
+    {%- endif %}
+{%- endif %}
+{%- set ns = namespace(multi_step_tool=true, last_query_index=messages|length - 1) %}
+{%- for message in messages[::-1] %}
+    {%- set index = (messages|length - 1) - loop.index0 %}
+    {%- if ns.multi_step_tool and message.role == "user" %}
+        {%- set content = render_content(message.content, false)|trim %}
+        {%- if not(content.startswith('<tool_response>') and content.endswith('</tool_response>')) %}
+            {%- set ns.multi_step_tool = false %}
+            {%- set ns.last_query_index = index %}
+        {%- endif %}
+    {%- endif %}
+{%- endfor %}
+{%- if ns.multi_step_tool %}
+    {{- raise_exception('No user query found in messages.') }}
+{%- endif %}
+{%- for message in messages %}
+    {%- set content = render_content(message.content, true)|trim %}
+    {%- if message.role == "system" %}
+        {%- if not loop.first %}
+            {{- raise_exception('System message must be at the beginning.') }}
+        {%- endif %}
+    {%- elif message.role == "user" %}
+        {{- '<|im_start|>' + message.role + '\n' + content + '<|im_end|>' + '\n' }}
+    {%- elif message.role == "assistant" %}
+        {%- set reasoning_content = '' %}
+        {%- if message.reasoning_content is string %}
+            {%- set reasoning_content = message.reasoning_content %}
+        {%- else %}
+            {%- if '</think>' in content %}
+                {%- set reasoning_content = content.split('</think>')[0].rstrip('\n').split('<think>')[-1].lstrip('\n') %}
+                {%- set content = content.split('</think>')[-1].lstrip('\n') %}
+            {%- endif %}
+        {%- endif %}
+        {%- set reasoning_content = reasoning_content|trim %}
+        {%- if loop.index0 > ns.last_query_index %}
+            {{- '<|im_start|>' + message.role + '\n<think>\n' + reasoning_content + '\n</think>\n\n' + content }}
+        {%- else %}
+            {{- '<|im_start|>' + message.role + '\n' + content }}
+        {%- endif %}
+        {%- if message.tool_calls and message.tool_calls is iterable and message.tool_calls is not mapping %}
+            {%- for tool_call in message.tool_calls %}
+                {%- if tool_call.function is defined %}
+                    {%- set tool_call = tool_call.function %}
+                {%- endif %}
+                {%- if loop.first %}
+                    {%- if content|trim %}
+                        {{- '\n\n<tool_call>\n<function=' + tool_call.name + '>\n' }}
+                    {%- else %}
+                        {{- '<tool_call>\n<function=' + tool_call.name + '>\n' }}
+                    {%- endif %}
+                {%- else %}
+                    {{- '\n<tool_call>\n<function=' + tool_call.name + '>\n' }}
+                {%- endif %}
+                {%- if tool_call.arguments is defined %}
+                    {%- for args_name, args_value in tool_call.arguments|items %}
+                        {{- '<parameter=' + args_name + '>\n' }}
+                        {%- set args_value = args_value | tojson | safe if args_value is mapping or (args_value is sequence and args_value is not string) else args_value | string %}
+                        {{- args_value }}
+                        {{- '\n</parameter>\n' }}
+                    {%- endfor %}
+                {%- endif %}
+                {{- '</function>\n</tool_call>' }}
+            {%- endfor %}
+        {%- endif %}
+        {{- '<|im_end|>\n' }}
+    {%- elif message.role == "tool" %}
+        {%- if loop.previtem and loop.previtem.role != "tool" %}
+            {{- '<|im_start|>user' }}
+        {%- endif %}
+        {{- '\n<tool_response>\n' }}
+        {{- content }}
+        {{- '\n</tool_response>' }}
+        {%- if not loop.last and loop.nextitem.role != "tool" %}
+            {{- '<|im_end|>\n' }}
+        {%- elif loop.last %}
+            {{- '<|im_end|>\n' }}
+        {%- endif %}
+    {%- else %}
+        {{- raise_exception('Unexpected message role.') }}
+    {%- endif %}
+{%- endfor %}
+{%- if add_generation_prompt %}
+    {{- '<|im_start|>assistant\n' }}
+    {%- if enable_thinking is defined and enable_thinking is false %}
+        {{- '<think>\n\n</think>\n\n' }}
+    {%- else %}
+        {{- '<think>\n' }}
+    {%- endif %}
+{%- endif %}

overgeneralisation_original_Estonian/Qwen3.5-4B-Base_overgeneralisation_splits_original_features_train_overgeneralisation_splits_original_features_test2/checkpoint-1100/tokenizer_config.json

@@ -0,0 +1,31 @@
+{
+  "add_prefix_space": false,
+  "audio_bos_token": "<|audio_start|>",
+  "audio_eos_token": "<|audio_end|>",
+  "audio_token": "<|audio_pad|>",
+  "backend": "tokenizers",
+  "bos_token": null,
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<|endoftext|>",
+  "errors": "replace",
+  "image_token": "<|image_pad|>",
+  "is_local": false,
+  "model_max_length": 262144,
+  "model_specific_special_tokens": {
+    "audio_bos_token": "<|audio_start|>",
+    "audio_eos_token": "<|audio_end|>",
+    "audio_token": "<|audio_pad|>",
+    "image_token": "<|image_pad|>",
+    "video_token": "<|video_pad|>",
+    "vision_bos_token": "<|vision_start|>",
+    "vision_eos_token": "<|vision_end|>"
+  },
+  "pad_token": "<|endoftext|>",
+  "pretokenize_regex": "(?i:'s|'t|'re|'ve|'m|'ll|'d)|[^\\r\\n\\p{L}\\p{N}]?[\\p{L}\\p{M}]+|\\p{N}| ?[^\\s\\p{L}\\p{M}\\p{N}]+[\\r\\n]*|\\s*[\\r\\n]+|\\s+(?!\\S)|\\s+",
+  "split_special_tokens": false,
+  "tokenizer_class": "TokenizersBackend",
+  "unk_token": null,
+  "video_token": "<|video_pad|>",
+  "vision_bos_token": "<|vision_start|>",
+  "vision_eos_token": "<|vision_end|>"
+}

overgeneralisation_original_Estonian/Qwen3.5-4B-Base_overgeneralisation_splits_original_features_train_overgeneralisation_splits_original_features_test2/checkpoint-1100/trainer_state.json

@@ -0,0 +1,1189 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 2.651780325890163,
+  "eval_steps": 20,
+  "global_step": 1100,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "entropy": 2.172999547421932,
+      "epoch": 0.04828002414001207,
+      "grad_norm": 2.5071208477020264,
+      "learning_rate": 1.7227585487708594e-05,
+      "loss": 1.9347333908081055,
+      "mean_token_accuracy": 0.6086816020309925,
+      "num_tokens": 47778.0,
+      "step": 20
+    },
+    {
+      "epoch": 0.04828002414001207,
+      "eval_entropy": 1.6909754791956269,
+      "eval_loss": 1.4728385210037231,
+      "eval_mean_token_accuracy": 0.6773593797442619,
+      "eval_num_tokens": 47778.0,
+      "eval_runtime": 90.1265,
+      "eval_samples_per_second": 15.756,
+      "eval_steps_per_second": 1.975,
+      "step": 20
+    },
+    {
+      "entropy": 1.2014626950025558,
+      "epoch": 0.09656004828002414,
+      "grad_norm": 1.5074543952941895,
+      "learning_rate": 3.5361886001086065e-05,
+      "loss": 1.0867146492004394,
+      "mean_token_accuracy": 0.7285927519202232,
+      "num_tokens": 95981.0,
+      "step": 40
+    },
+    {
+      "epoch": 0.09656004828002414,
+      "eval_entropy": 1.0304168352250302,
+      "eval_loss": 0.9211422204971313,
+      "eval_mean_token_accuracy": 0.7613122848312507,
+      "eval_num_tokens": 95981.0,
+      "eval_runtime": 88.6338,
+      "eval_samples_per_second": 16.021,
+      "eval_steps_per_second": 2.008,
+      "step": 40
+    },
+    {
+      "entropy": 0.9454198732972146,
+      "epoch": 0.14484007242003621,
+      "grad_norm": 1.7578108310699463,
+      "learning_rate": 5.349618651446354e-05,
+      "loss": 0.8447297096252442,
+      "mean_token_accuracy": 0.7723424732685089,
+      "num_tokens": 142784.0,
+      "step": 60
+    },
+    {
+      "epoch": 0.14484007242003621,
+      "eval_entropy": 0.8647834474451086,
+      "eval_loss": 0.8240737318992615,
+      "eval_mean_token_accuracy": 0.7821227014064789,
+      "eval_num_tokens": 142784.0,
+      "eval_runtime": 88.0135,
+      "eval_samples_per_second": 16.134,
+      "eval_steps_per_second": 2.022,
+      "step": 60
+    },
+    {
+      "entropy": 0.8963550612330436,
+      "epoch": 0.19312009656004828,
+      "grad_norm": 1.1464931964874268,
+      "learning_rate": 7.1630487027841e-05,
+      "loss": 0.7979836463928223,
+      "mean_token_accuracy": 0.7822489373385906,
+      "num_tokens": 187338.0,
+      "step": 80
+    },
+    {
+      "epoch": 0.19312009656004828,
+      "eval_entropy": 0.8663296133614657,
+      "eval_loss": 0.7752988338470459,
+      "eval_mean_token_accuracy": 0.7885957012015782,
+      "eval_num_tokens": 187338.0,
+      "eval_runtime": 88.284,
+      "eval_samples_per_second": 16.084,
+      "eval_steps_per_second": 2.016,
+      "step": 80
+    },
+    {
+      "entropy": 0.8265103206038475,
+      "epoch": 0.24140012070006034,
+      "grad_norm": 1.0106662511825562,
+      "learning_rate": 8.976478754121848e-05,
+      "loss": 0.7297060012817382,
+      "mean_token_accuracy": 0.7955138415098191,
+      "num_tokens": 236038.0,
+      "step": 100
+    },
+    {
+      "epoch": 0.24140012070006034,
+      "eval_entropy": 0.8122169536151244,
+      "eval_loss": 0.754906177520752,
+      "eval_mean_token_accuracy": 0.7917155255092664,
+      "eval_num_tokens": 236038.0,
+      "eval_runtime": 87.5754,
+      "eval_samples_per_second": 16.215,
+      "eval_steps_per_second": 2.033,
+      "step": 100
+    },
+    {
+      "entropy": 0.8098392844200134,
+      "epoch": 0.28968014484007243,
+      "grad_norm": 1.2238041162490845,
+      "learning_rate": 0.00010789908805459595,
+      "loss": 0.7201489448547364,
+      "mean_token_accuracy": 0.7975016921758652,
+      "num_tokens": 282918.0,
+      "step": 120
+    },
+    {
+      "epoch": 0.28968014484007243,
+      "eval_entropy": 0.8418726462326692,
+      "eval_loss": 0.7393178939819336,
+      "eval_mean_token_accuracy": 0.7936771585700217,
+      "eval_num_tokens": 282918.0,
+      "eval_runtime": 87.1597,
+      "eval_samples_per_second": 16.292,
+      "eval_steps_per_second": 2.042,
+      "step": 120
+    },
+    {
+      "entropy": 0.8080591082572937,
+      "epoch": 0.33796016898008446,
+      "grad_norm": 1.011591911315918,
+      "learning_rate": 0.00012603338856797343,
+      "loss": 0.7147697925567627,
+      "mean_token_accuracy": 0.7947175532579422,
+      "num_tokens": 331305.0,
+      "step": 140
+    },
+    {
+      "epoch": 0.33796016898008446,
+      "eval_entropy": 0.7894941431083037,
+      "eval_loss": 0.720219075679779,
+      "eval_mean_token_accuracy": 0.801899236239744,
+      "eval_num_tokens": 331305.0,
+      "eval_runtime": 87.1493,
+      "eval_samples_per_second": 16.294,
+      "eval_steps_per_second": 2.042,
+      "step": 140
+    },
+    {
+      "entropy": 0.7957205109298229,
+      "epoch": 0.38624019312009655,
+      "grad_norm": 0.8481830954551697,
+      "learning_rate": 0.00014416768908135088,
+      "loss": 0.7050027847290039,
+      "mean_token_accuracy": 0.7995368830859662,
+      "num_tokens": 375876.0,
+      "step": 160
+    },
+    {
+      "epoch": 0.38624019312009655,
+      "eval_entropy": 0.7746140591883928,
+      "eval_loss": 0.7175475358963013,
+      "eval_mean_token_accuracy": 0.7994848955213354,
+      "eval_num_tokens": 375876.0,
+      "eval_runtime": 86.655,
+      "eval_samples_per_second": 16.387,
+      "eval_steps_per_second": 2.054,
+      "step": 160
+    },
+    {
+      "entropy": 0.779565304517746,
+      "epoch": 0.43452021726010864,
+      "grad_norm": 1.032624363899231,
+      "learning_rate": 0.00016230198959472835,
+      "loss": 0.6978332042694092,
+      "mean_token_accuracy": 0.8015618294477462,
+      "num_tokens": 423913.0,
+      "step": 180
+    },
+    {
+      "epoch": 0.43452021726010864,
+      "eval_entropy": 0.7455583871080634,
+      "eval_loss": 0.7118850350379944,
+      "eval_mean_token_accuracy": 0.7999682195400923,
+      "eval_num_tokens": 423913.0,
+      "eval_runtime": 86.7842,
+      "eval_samples_per_second": 16.362,
+      "eval_steps_per_second": 2.051,
+      "step": 180
+    },
+    {
+      "entropy": 0.7810172818601131,
+      "epoch": 0.4828002414001207,
+      "grad_norm": 0.9276663064956665,
+      "learning_rate": 0.00018043629010810582,
+      "loss": 0.6937861442565918,
+      "mean_token_accuracy": 0.802893053740263,
+      "num_tokens": 472395.0,
+      "step": 200
+    },
+    {
+      "epoch": 0.4828002414001207,
+      "eval_entropy": 0.7650193290764027,
+      "eval_loss": 0.7020567059516907,
+      "eval_mean_token_accuracy": 0.8045302153973097,
+      "eval_num_tokens": 472395.0,
+      "eval_runtime": 88.326,
+      "eval_samples_per_second": 16.077,
+      "eval_steps_per_second": 2.015,
+      "step": 200
+    },
+    {
+      "entropy": 0.7862060204148292,
+      "epoch": 0.5310802655401328,
+      "grad_norm": 0.8195295929908752,
+      "learning_rate": 0.0001985705906214833,
+      "loss": 0.6878782272338867,
+      "mean_token_accuracy": 0.8027168907225132,
+      "num_tokens": 517049.0,
+      "step": 220
+    },
+    {
+      "epoch": 0.5310802655401328,
+      "eval_entropy": 0.7656565583154057,
+      "eval_loss": 0.6973932981491089,
+      "eval_mean_token_accuracy": 0.8047967278555538,
+      "eval_num_tokens": 517049.0,
+      "eval_runtime": 87.7698,
+      "eval_samples_per_second": 16.179,
+      "eval_steps_per_second": 2.028,
+      "step": 220
+    },
+    {
+      "entropy": 0.7611534461379051,
+      "epoch": 0.5793602896801449,
+      "grad_norm": 0.7499712109565735,
+      "learning_rate": 0.00021670489113486077,
+      "loss": 0.681937837600708,
+      "mean_token_accuracy": 0.8041978515684605,
+      "num_tokens": 562836.0,
+      "step": 240
+    },
+    {
+      "epoch": 0.5793602896801449,
+      "eval_entropy": 0.7599442380197933,
+      "eval_loss": 0.7002046704292297,
+      "eval_mean_token_accuracy": 0.8051861517884759,
+      "eval_num_tokens": 562836.0,
+      "eval_runtime": 87.9239,
+      "eval_samples_per_second": 16.15,
+      "eval_steps_per_second": 2.024,
+      "step": 240
+    },
+    {
+      "entropy": 0.7800783462822437,
+      "epoch": 0.627640313820157,
+      "grad_norm": 0.8645269274711609,
+      "learning_rate": 0.00023483919164823825,
+      "loss": 0.6954500675201416,
+      "mean_token_accuracy": 0.8031484372913837,
+      "num_tokens": 606612.0,
+      "step": 260
+    },
+    {
+      "epoch": 0.627640313820157,
+      "eval_entropy": 0.7498679090751691,
+      "eval_loss": 0.7046149373054504,
+      "eval_mean_token_accuracy": 0.802388215667746,
+      "eval_num_tokens": 606612.0,
+      "eval_runtime": 88.2468,
+      "eval_samples_per_second": 16.091,
+      "eval_steps_per_second": 2.017,
+      "step": 260
+    },
+    {
+      "entropy": 0.7586533665657044,
+      "epoch": 0.6759203379601689,
+      "grad_norm": 0.741596519947052,
+      "learning_rate": 0.0002529734921616157,
+      "loss": 0.690476369857788,
+      "mean_token_accuracy": 0.8046923100948333,
+      "num_tokens": 655920.0,
+      "step": 280
+    },
+    {
+      "epoch": 0.6759203379601689,
+      "eval_entropy": 0.7714274066887544,
+      "eval_loss": 0.6992160677909851,
+      "eval_mean_token_accuracy": 0.8042646482419432,
+      "eval_num_tokens": 655920.0,
+      "eval_runtime": 88.1754,
+      "eval_samples_per_second": 16.104,
+      "eval_steps_per_second": 2.019,
+      "step": 280
+    },
+    {
+      "entropy": 0.7585848286747933,
+      "epoch": 0.724200362100181,
+      "grad_norm": 1.0778124332427979,
+      "learning_rate": 0.00027110779267499317,
+      "loss": 0.6790409564971924,
+      "mean_token_accuracy": 0.8060351841151714,
+      "num_tokens": 702638.0,
+      "step": 300
+    },
+    {
+      "epoch": 0.724200362100181,
+      "eval_entropy": 0.7380402811457601,
+      "eval_loss": 0.7054820656776428,
+      "eval_mean_token_accuracy": 0.8003534738267406,
+      "eval_num_tokens": 702638.0,
+      "eval_runtime": 88.238,
+      "eval_samples_per_second": 16.093,
+      "eval_steps_per_second": 2.017,
+      "step": 300
+    },
+    {
+      "entropy": 0.7696981988847256,
+      "epoch": 0.7724803862401931,
+      "grad_norm": 0.8408658504486084,
+      "learning_rate": 0.00028924209318837064,
+      "loss": 0.7070387363433838,
+      "mean_token_accuracy": 0.8035947173833847,
+      "num_tokens": 749377.0,
+      "step": 320
+    },
+    {
+      "epoch": 0.7724803862401931,
+      "eval_entropy": 0.7897409023193831,
+      "eval_loss": 0.7015668749809265,
+      "eval_mean_token_accuracy": 0.805306687113944,
+      "eval_num_tokens": 749377.0,
+      "eval_runtime": 87.9395,
+      "eval_samples_per_second": 16.147,
+      "eval_steps_per_second": 2.024,
+      "step": 320
+    },
+    {
+      "entropy": 0.762337189912796,
+      "epoch": 0.8207604103802052,
+      "grad_norm": 0.9843310117721558,
+      "learning_rate": 0.0003073763937017481,
+      "loss": 0.7021088123321533,
+      "mean_token_accuracy": 0.8029616877436638,
+      "num_tokens": 798874.0,
+      "step": 340
+    },
+    {
+      "epoch": 0.8207604103802052,
+      "eval_entropy": 0.7210077121016685,
+      "eval_loss": 0.7082746624946594,
+      "eval_mean_token_accuracy": 0.8049283824609906,
+      "eval_num_tokens": 798874.0,
+      "eval_runtime": 88.1706,
+      "eval_samples_per_second": 16.105,
+      "eval_steps_per_second": 2.019,
+      "step": 340
+    },
+    {
+      "entropy": 0.7879025422036647,
+      "epoch": 0.8690404345202173,
+      "grad_norm": 0.9713814854621887,
+      "learning_rate": 0.0003255106942151256,
+      "loss": 0.714624547958374,
+      "mean_token_accuracy": 0.79987031519413,
+      "num_tokens": 840825.0,
+      "step": 360
+    },
+    {
+      "epoch": 0.8690404345202173,
+      "eval_entropy": 0.7866930489459735,
+      "eval_loss": 0.7145519852638245,
+      "eval_mean_token_accuracy": 0.8015657988157165,
+      "eval_num_tokens": 840825.0,
+      "eval_runtime": 88.3746,
+      "eval_samples_per_second": 16.068,
+      "eval_steps_per_second": 2.014,
+      "step": 360
+    },
+    {
+      "entropy": 0.7844726584851742,
+      "epoch": 0.9173204586602294,
+      "grad_norm": 1.2028311491012573,
+      "learning_rate": 0.00034364499472850306,
+      "loss": 0.7040786266326904,
+      "mean_token_accuracy": 0.7981885217130185,
+      "num_tokens": 882885.0,
+      "step": 380
+    },
+    {
+      "epoch": 0.9173204586602294,
+      "eval_entropy": 0.7767668339643585,
+      "eval_loss": 0.7073574066162109,
+      "eval_mean_token_accuracy": 0.8040851833445303,
+      "eval_num_tokens": 882885.0,
+      "eval_runtime": 88.3828,
+      "eval_samples_per_second": 16.066,
+      "eval_steps_per_second": 2.014,
+      "step": 380
+    },
+    {
+      "entropy": 0.7699790760874748,
+      "epoch": 0.9656004828002414,
+      "grad_norm": 0.9397398233413696,
+      "learning_rate": 0.00036177929524188054,
+      "loss": 0.7040313720703125,
+      "mean_token_accuracy": 0.8034727744758129,
+      "num_tokens": 927456.0,
+      "step": 400
+    },
+    {
+      "epoch": 0.9656004828002414,
+      "eval_entropy": 0.7749103545472863,
+      "eval_loss": 0.7093513011932373,
+      "eval_mean_token_accuracy": 0.8019482901926791,
+      "eval_num_tokens": 927456.0,
+      "eval_runtime": 86.2112,
+      "eval_samples_per_second": 16.471,
+      "eval_steps_per_second": 2.065,
+      "step": 400
+    },
+    {
+      "entropy": 0.7672389172888422,
+      "epoch": 1.012070006035003,
+      "grad_norm": 0.99210524559021,
+      "learning_rate": 0.00037628567078152296,
+      "loss": 0.6913118362426758,
+      "mean_token_accuracy": 0.8042210944287189,
+      "num_tokens": 972662.0,
+      "step": 420
+    },
+    {
+      "epoch": 1.012070006035003,
+      "eval_entropy": 0.7294771229283193,
+      "eval_loss": 0.7171520590782166,
+      "eval_mean_token_accuracy": 0.8015128585059991,
+      "eval_num_tokens": 972662.0,
+      "eval_runtime": 88.037,
+      "eval_samples_per_second": 16.13,
+      "eval_steps_per_second": 2.022,
+      "step": 420
+    },
+    {
+      "entropy": 0.7143943183124065,
+      "epoch": 1.060350030175015,
+      "grad_norm": 1.087638258934021,
+      "learning_rate": 0.0003762484015601069,
+      "loss": 0.642704439163208,
+      "mean_token_accuracy": 0.8124966286122799,
+      "num_tokens": 1022525.0,
+      "step": 440
+    },
+    {
+      "epoch": 1.060350030175015,
+      "eval_entropy": 0.743511886744017,
+      "eval_loss": 0.7116662859916687,
+      "eval_mean_token_accuracy": 0.8037210672758939,
+      "eval_num_tokens": 1022525.0,
+      "eval_runtime": 87.9931,
+      "eval_samples_per_second": 16.138,
+      "eval_steps_per_second": 2.023,
+      "step": 440
+    },
+    {
+      "entropy": 0.7237346574664116,
+      "epoch": 1.1086300543150271,
+      "grad_norm": 0.9365553259849548,
+      "learning_rate": 0.0003761579008392123,
+      "loss": 0.652644681930542,
+      "mean_token_accuracy": 0.8107496216893196,
+      "num_tokens": 1069194.0,
+      "step": 460
+    },
+    {
+      "epoch": 1.1086300543150271,
+      "eval_entropy": 0.7372823839776972,
+      "eval_loss": 0.7238303422927856,
+      "eval_mean_token_accuracy": 0.8026230124275336,
+      "eval_num_tokens": 1069194.0,
+      "eval_runtime": 87.6778,
+      "eval_samples_per_second": 16.196,
+      "eval_steps_per_second": 2.03,
+      "step": 460
+    },
+    {
+      "entropy": 0.7202137842774391,
+      "epoch": 1.1569100784550392,
+      "grad_norm": 0.7998443841934204,
+      "learning_rate": 0.0003760141942294726,
+      "loss": 0.6485219478607178,
+      "mean_token_accuracy": 0.8092714451253414,
+      "num_tokens": 1117570.0,
+      "step": 480
+    },
+    {
+      "epoch": 1.1569100784550392,
+      "eval_entropy": 0.71547647942318,
+      "eval_loss": 0.7124277949333191,
+      "eval_mean_token_accuracy": 0.8013174219077892,
+      "eval_num_tokens": 1117570.0,
+      "eval_runtime": 87.4387,
+      "eval_samples_per_second": 16.24,
+      "eval_steps_per_second": 2.036,
+      "step": 480
+    },
+    {
+      "entropy": 0.7270819500088692,
+      "epoch": 1.2051901025950513,
+      "grad_norm": 0.9470787644386292,
+      "learning_rate": 0.00037581732239815854,
+      "loss": 0.66352219581604,
+      "mean_token_accuracy": 0.8102593503892421,
+      "num_tokens": 1161717.0,
+      "step": 500
+    },
+    {
+      "epoch": 1.2051901025950513,
+      "eval_entropy": 0.7791565897759427,
+      "eval_loss": 0.7179726958274841,
+      "eval_mean_token_accuracy": 0.7948898328154275,
+      "eval_num_tokens": 1161717.0,
+      "eval_runtime": 87.9927,
+      "eval_samples_per_second": 16.138,
+      "eval_steps_per_second": 2.023,
+      "step": 500
+    },
+    {
+      "entropy": 0.7320936664938926,
+      "epoch": 1.2534701267350634,
+      "grad_norm": 1.0220379829406738,
+      "learning_rate": 0.0003755673410576695,
+      "loss": 0.6581549644470215,
+      "mean_token_accuracy": 0.8098709337413311,
+      "num_tokens": 1209390.0,
+      "step": 520
+    },
+    {
+      "epoch": 1.2534701267350634,
+      "eval_entropy": 0.7710678225822662,
+      "eval_loss": 0.708797812461853,
+      "eval_mean_token_accuracy": 0.8039572778042783,
+      "eval_num_tokens": 1209390.0,
+      "eval_runtime": 88.1193,
+      "eval_samples_per_second": 16.115,
+      "eval_steps_per_second": 2.02,
+      "step": 520
+    },
+    {
+      "entropy": 0.7281674664467573,
+      "epoch": 1.3017501508750755,
+      "grad_norm": 0.8775188326835632,
+      "learning_rate": 0.000375264320949768,
+      "loss": 0.6631278991699219,
+      "mean_token_accuracy": 0.8101261422038079,
+      "num_tokens": 1256611.0,
+      "step": 540
+    },
+    {
+      "epoch": 1.3017501508750755,
+      "eval_entropy": 0.7199799800187014,
+      "eval_loss": 0.6981531381607056,
+      "eval_mean_token_accuracy": 0.8071265254127845,
+      "eval_num_tokens": 1256611.0,
+      "eval_runtime": 88.165,
+      "eval_samples_per_second": 16.106,
+      "eval_steps_per_second": 2.019,
+      "step": 540
+    },
+    {
+      "entropy": 0.722860773652792,
+      "epoch": 1.3500301750150876,
+      "grad_norm": 0.8949288129806519,
+      "learning_rate": 0.00037490834782556,
+      "loss": 0.6639598846435547,
+      "mean_token_accuracy": 0.8105649061501026,
+      "num_tokens": 1302325.0,
+      "step": 560
+    },
+    {
+      "epoch": 1.3500301750150876,
+      "eval_entropy": 0.7134760032209118,
+      "eval_loss": 0.7029948234558105,
+      "eval_mean_token_accuracy": 0.8062985880991046,
+      "eval_num_tokens": 1302325.0,
+      "eval_runtime": 88.1286,
+      "eval_samples_per_second": 16.113,
+      "eval_steps_per_second": 2.02,
+      "step": 560
+    },
+    {
+      "entropy": 0.7299126699566841,
+      "epoch": 1.3983101991550995,
+      "grad_norm": 1.0323039293289185,
+      "learning_rate": 0.0003744995224212291,
+      "loss": 0.6535763263702392,
+      "mean_token_accuracy": 0.8120224848389626,
+      "num_tokens": 1343844.0,
+      "step": 580
+    },
+    {
+      "epoch": 1.3983101991550995,
+      "eval_entropy": 0.7169701182440426,
+      "eval_loss": 0.7067857980728149,
+      "eval_mean_token_accuracy": 0.8040432936689826,
+      "eval_num_tokens": 1343844.0,
+      "eval_runtime": 88.2252,
+      "eval_samples_per_second": 16.095,
+      "eval_steps_per_second": 2.018,
+      "step": 580
+    },
+    {
+      "entropy": 0.7281742256134749,
+      "epoch": 1.4465902232951118,
+      "grad_norm": 1.3370424509048462,
+      "learning_rate": 0.00037403796042952863,
+      "loss": 0.6637272357940673,
+      "mean_token_accuracy": 0.8118261776864528,
+      "num_tokens": 1389471.0,
+      "step": 600
+    },
+    {
+      "epoch": 1.4465902232951118,
+      "eval_entropy": 0.7378843212395572,
+      "eval_loss": 0.7038390040397644,
+      "eval_mean_token_accuracy": 0.8059229371922739,
+      "eval_num_tokens": 1389471.0,
+      "eval_runtime": 87.9301,
+      "eval_samples_per_second": 16.149,
+      "eval_steps_per_second": 2.024,
+      "step": 600
+    },
+    {
+      "entropy": 0.7388614989817143,
+      "epoch": 1.4948702474351236,
+      "grad_norm": 0.8464171290397644,
+      "learning_rate": 0.00037352379246704257,
+      "loss": 0.6604740142822265,
+      "mean_token_accuracy": 0.8102974124252796,
+      "num_tokens": 1435975.0,
+      "step": 620
+    },
+    {
+      "epoch": 1.4948702474351236,
+      "eval_entropy": 0.7126847046814607,
+      "eval_loss": 0.6950703263282776,
+      "eval_mean_token_accuracy": 0.8085885392815879,
+      "eval_num_tokens": 1435975.0,
+      "eval_runtime": 88.2732,
+      "eval_samples_per_second": 16.086,
+      "eval_steps_per_second": 2.016,
+      "step": 620
+    },
+    {
+      "entropy": 0.7258916139602661,
+      "epoch": 1.5431502715751357,
+      "grad_norm": 1.1780085563659668,
+      "learning_rate": 0.0003729571640372223,
+      "loss": 0.6529891014099121,
+      "mean_token_accuracy": 0.8085981778800487,
+      "num_tokens": 1482035.0,
+      "step": 640
+    },
+    {
+      "epoch": 1.5431502715751357,
+      "eval_entropy": 0.7096509638797032,
+      "eval_loss": 0.690051257610321,
+      "eval_mean_token_accuracy": 0.8100520241796301,
+      "eval_num_tokens": 1482035.0,
+      "eval_runtime": 88.3153,
+      "eval_samples_per_second": 16.079,
+      "eval_steps_per_second": 2.016,
+      "step": 640
+    },
+    {
+      "entropy": 0.7151173129677773,
+      "epoch": 1.5914302957151478,
+      "grad_norm": 1.0357667207717896,
+      "learning_rate": 0.0003723382354892108,
+      "loss": 0.6451474189758301,
+      "mean_token_accuracy": 0.8136902332305909,
+      "num_tokens": 1528119.0,
+      "step": 660
+    },
+    {
+      "epoch": 1.5914302957151478,
+      "eval_entropy": 0.7060252652409371,
+      "eval_loss": 0.6899483799934387,
+      "eval_mean_token_accuracy": 0.810834194502134,
+      "eval_num_tokens": 1528119.0,
+      "eval_runtime": 87.8848,
+      "eval_samples_per_second": 16.158,
+      "eval_steps_per_second": 2.025,
+      "step": 660
+    },
+    {
+      "entropy": 0.7141377851366997,
+      "epoch": 1.63971031985516,
+      "grad_norm": 0.8391667008399963,
+      "learning_rate": 0.000371667181972466,
+      "loss": 0.6481579780578614,
+      "mean_token_accuracy": 0.8159015104174614,
+      "num_tokens": 1573661.0,
+      "step": 680
+    },
+    {
+      "epoch": 1.63971031985516,
+      "eval_entropy": 0.7598993319473909,
+      "eval_loss": 0.681204080581665,
+      "eval_mean_token_accuracy": 0.8100194586126992,
+      "eval_num_tokens": 1573661.0,
+      "eval_runtime": 88.2415,
+      "eval_samples_per_second": 16.092,
+      "eval_steps_per_second": 2.017,
+      "step": 680
+    },
+    {
+      "entropy": 0.7235980287194252,
+      "epoch": 1.687990343995172,
+      "grad_norm": 0.8197196125984192,
+      "learning_rate": 0.00037094419338719537,
+      "loss": 0.6582849025726318,
+      "mean_token_accuracy": 0.8114834539592266,
+      "num_tokens": 1621714.0,
+      "step": 700
+    },
+    {
+      "epoch": 1.687990343995172,
+      "eval_entropy": 0.7166854568411795,
+      "eval_loss": 0.679177463054657,
+      "eval_mean_token_accuracy": 0.8119039877077167,
+      "eval_num_tokens": 1621714.0,
+      "eval_runtime": 88.0957,
+      "eval_samples_per_second": 16.119,
+      "eval_steps_per_second": 2.021,
+      "step": 700
+    },
+    {
+      "entropy": 0.7278190158307553,
+      "epoch": 1.736270368135184,
+      "grad_norm": 0.7820602655410767,
+      "learning_rate": 0.0003701694743306164,
+      "loss": 0.6633153438568116,
+      "mean_token_accuracy": 0.8123249113559723,
+      "num_tokens": 1672118.0,
+      "step": 720
+    },
+    {
+      "epoch": 1.736270368135184,
+      "eval_entropy": 0.6980850314826108,
+      "eval_loss": 0.6775653958320618,
+      "eval_mean_token_accuracy": 0.8116872906684875,
+      "eval_num_tokens": 1672118.0,
+      "eval_runtime": 87.9795,
+      "eval_samples_per_second": 16.14,
+      "eval_steps_per_second": 2.023,
+      "step": 720
+    },
+    {
+      "entropy": 0.7149631556123495,
+      "epoch": 1.7845503922751962,
+      "grad_norm": 1.0139760971069336,
+      "learning_rate": 0.00036934324403905795,
+      "loss": 0.6468976020812989,
+      "mean_token_accuracy": 0.8117561548948288,
+      "num_tokens": 1720465.0,
+      "step": 740
+    },
+    {
+      "epoch": 1.7845503922751962,
+      "eval_entropy": 0.69009849094273,
+      "eval_loss": 0.6731093525886536,
+      "eval_mean_token_accuracy": 0.81281964196248,
+      "eval_num_tokens": 1720465.0,
+      "eval_runtime": 87.7412,
+      "eval_samples_per_second": 16.184,
+      "eval_steps_per_second": 2.029,
+      "step": 740
+    },
+    {
+      "entropy": 0.7226049453020096,
+      "epoch": 1.832830416415208,
+      "grad_norm": 1.5040546655654907,
+      "learning_rate": 0.0003684657363259193,
+      "loss": 0.6385328769683838,
+      "mean_token_accuracy": 0.8132639616727829,
+      "num_tokens": 1764440.0,
+      "step": 760
+    },
+    {
+      "epoch": 1.832830416415208,
+      "eval_entropy": 0.7125169618076153,
+      "eval_loss": 0.6864680647850037,
+      "eval_mean_token_accuracy": 0.8049418089095126,
+      "eval_num_tokens": 1764440.0,
+      "eval_runtime": 87.9429,
+      "eval_samples_per_second": 16.147,
+      "eval_steps_per_second": 2.024,
+      "step": 760
+    },
+    {
+      "entropy": 0.707821810990572,
+      "epoch": 1.8811104405552204,
+      "grad_norm": 0.8604223132133484,
+      "learning_rate": 0.00036753719951550327,
+      "loss": 0.6397049427032471,
+      "mean_token_accuracy": 0.8137877315282822,
+      "num_tokens": 1810828.0,
+      "step": 780
+    },
+    {
+      "epoch": 1.8811104405552204,
+      "eval_entropy": 0.6492102551326323,
+      "eval_loss": 0.6743726134300232,
+      "eval_mean_token_accuracy": 0.8130356572317273,
+      "eval_num_tokens": 1810828.0,
+      "eval_runtime": 87.8684,
+      "eval_samples_per_second": 16.161,
+      "eval_steps_per_second": 2.026,
+      "step": 780
+    },
+    {
+      "entropy": 0.6800355084240437,
+      "epoch": 1.9293904646952322,
+      "grad_norm": 0.7270982265472412,
+      "learning_rate": 0.00036655789637274377,
+      "loss": 0.6267755031585693,
+      "mean_token_accuracy": 0.8180945307016373,
+      "num_tokens": 1857600.0,
+      "step": 800
+    },
+    {
+      "epoch": 1.9293904646952322,
+      "eval_entropy": 0.6727538651294922,
+      "eval_loss": 0.6696028709411621,
+      "eval_mean_token_accuracy": 0.8116443090224534,
+      "eval_num_tokens": 1857600.0,
+      "eval_runtime": 88.2449,
+      "eval_samples_per_second": 16.092,
+      "eval_steps_per_second": 2.017,
+      "step": 800
+    },
+    {
+      "entropy": 0.6746691606938839,
+      "epoch": 1.9776704888352445,
+      "grad_norm": 0.9771540760993958,
+      "learning_rate": 0.0003655281040288461,
+      "loss": 0.6326992988586426,
+      "mean_token_accuracy": 0.8179976396262646,
+      "num_tokens": 1901998.0,
+      "step": 820
+    },
+    {
+      "epoch": 1.9776704888352445,
+      "eval_entropy": 0.6636319716324967,
+      "eval_loss": 0.6714363098144531,
+      "eval_mean_token_accuracy": 0.8126635943235976,
+      "eval_num_tokens": 1901998.0,
+      "eval_runtime": 87.7666,
+      "eval_samples_per_second": 16.179,
+      "eval_steps_per_second": 2.028,
+      "step": 820
+    },
+    {
+      "entropy": 0.6319125778489298,
+      "epoch": 2.024140012070006,
+      "grad_norm": 0.8380444645881653,
+      "learning_rate": 0.0003644481139028622,
+      "loss": 0.5780903816223144,
+      "mean_token_accuracy": 0.8250188401767186,
+      "num_tokens": 1947553.0,
+      "step": 840
+    },
+    {
+      "epoch": 2.024140012070006,
+      "eval_entropy": 0.6627151989535,
+      "eval_loss": 0.6774935722351074,
+      "eval_mean_token_accuracy": 0.8140510819601209,
+      "eval_num_tokens": 1947553.0,
+      "eval_runtime": 87.8287,
+      "eval_samples_per_second": 16.168,
+      "eval_steps_per_second": 2.027,
+      "step": 840
+    },
+    {
+      "entropy": 0.6059147048741579,
+      "epoch": 2.0724200362100182,
+      "grad_norm": 0.9745152592658997,
+      "learning_rate": 0.0003633182316192229,
+      "loss": 0.546702527999878,
+      "mean_token_accuracy": 0.8358186542987823,
+      "num_tokens": 1995234.0,
+      "step": 860
+    },
+    {
+      "epoch": 2.0724200362100182,
+      "eval_entropy": 0.6620710384979677,
+      "eval_loss": 0.6769291162490845,
+      "eval_mean_token_accuracy": 0.8135082835561773,
+      "eval_num_tokens": 1995234.0,
+      "eval_runtime": 88.0008,
+      "eval_samples_per_second": 16.136,
+      "eval_steps_per_second": 2.023,
+      "step": 860
+    },
+    {
+      "entropy": 0.6076117843389511,
+      "epoch": 2.12070006035003,
+      "grad_norm": 0.6914380192756653,
+      "learning_rate": 0.0003621387769212491,
+      "loss": 0.5324535846710206,
+      "mean_token_accuracy": 0.8366568259894848,
+      "num_tokens": 2040639.0,
+      "step": 880
+    },
+    {
+      "epoch": 2.12070006035003,
+      "eval_entropy": 0.6592224213514435,
+      "eval_loss": 0.670383632183075,
+      "eval_mean_token_accuracy": 0.8162150088320957,
+      "eval_num_tokens": 2040639.0,
+      "eval_runtime": 88.324,
+      "eval_samples_per_second": 16.077,
+      "eval_steps_per_second": 2.015,
+      "step": 880
+    },
+    {
+      "entropy": 0.6158072650432587,
+      "epoch": 2.1689800844900424,
+      "grad_norm": 0.9649198055267334,
+      "learning_rate": 0.0003609100835806688,
+      "loss": 0.5564568996429443,
+      "mean_token_accuracy": 0.8329654954373836,
+      "num_tokens": 2090604.0,
+      "step": 900
+    },
+    {
+      "epoch": 2.1689800844900424,
+      "eval_entropy": 0.608588819758276,
+      "eval_loss": 0.6745102405548096,
+      "eval_mean_token_accuracy": 0.8155080542135774,
+      "eval_num_tokens": 2090604.0,
+      "eval_runtime": 87.52,
+      "eval_samples_per_second": 16.225,
+      "eval_steps_per_second": 2.034,
+      "step": 900
+    },
+    {
+      "entropy": 0.5938734326511621,
+      "epoch": 2.2172601086300543,
+      "grad_norm": 0.6915557384490967,
+      "learning_rate": 0.0003596324993031632,
+      "loss": 0.5351875782012939,
+      "mean_token_accuracy": 0.8385631121695042,
+      "num_tokens": 2142246.0,
+      "step": 920
+    },
+    {
+      "epoch": 2.2172601086300543,
+      "eval_entropy": 0.6251631205001574,
+      "eval_loss": 0.6664518117904663,
+      "eval_mean_token_accuracy": 0.81711940417129,
+      "eval_num_tokens": 2142246.0,
+      "eval_runtime": 87.1792,
+      "eval_samples_per_second": 16.288,
+      "eval_steps_per_second": 2.042,
+      "step": 920
+    },
+    {
+      "entropy": 0.623173613846302,
+      "epoch": 2.2655401327700666,
+      "grad_norm": 0.8366610407829285,
+      "learning_rate": 0.00035830638562997063,
+      "loss": 0.5488213539123535,
+      "mean_token_accuracy": 0.8321483485400677,
+      "num_tokens": 2183144.0,
+      "step": 940
+    },
+    {
+      "epoch": 2.2655401327700666,
+      "eval_entropy": 0.6237085649136747,
+      "eval_loss": 0.68075031042099,
+      "eval_mean_token_accuracy": 0.814671738429016,
+      "eval_num_tokens": 2183144.0,
+      "eval_runtime": 88.0652,
+      "eval_samples_per_second": 16.124,
+      "eval_steps_per_second": 2.021,
+      "step": 940
+    },
+    {
+      "entropy": 0.6236911740154027,
+      "epoch": 2.3138201569100785,
+      "grad_norm": 0.7812452912330627,
+      "learning_rate": 0.00035693211783557416,
+      "loss": 0.5696200847625732,
+      "mean_token_accuracy": 0.82991396561265,
+      "num_tokens": 2228098.0,
+      "step": 960
+    },
+    {
+      "epoch": 2.3138201569100785,
+      "eval_entropy": 0.6327033426319615,
+      "eval_loss": 0.6742984652519226,
+      "eval_mean_token_accuracy": 0.8148086091105857,
+      "eval_num_tokens": 2228098.0,
+      "eval_runtime": 88.1711,
+      "eval_samples_per_second": 16.105,
+      "eval_steps_per_second": 2.019,
+      "step": 960
+    },
+    {
+      "entropy": 0.6218426413834095,
+      "epoch": 2.3621001810500903,
+      "grad_norm": 0.8719790577888489,
+      "learning_rate": 0.0003555100848215035,
+      "loss": 0.5582651615142822,
+      "mean_token_accuracy": 0.8323961742222309,
+      "num_tokens": 2273062.0,
+      "step": 980
+    },
+    {
+      "epoch": 2.3621001810500903,
+      "eval_entropy": 0.6338012375858393,
+      "eval_loss": 0.6685822606086731,
+      "eval_mean_token_accuracy": 0.8152180300669724,
+      "eval_num_tokens": 2273062.0,
+      "eval_runtime": 88.101,
+      "eval_samples_per_second": 16.118,
+      "eval_steps_per_second": 2.02,
+      "step": 980
+    },
+    {
+      "entropy": 0.6050681818276644,
+      "epoch": 2.4103802051901027,
+      "grad_norm": 0.9151566028594971,
+      "learning_rate": 0.00035404068900628076,
+      "loss": 0.559452486038208,
+      "mean_token_accuracy": 0.8310242861509323,
+      "num_tokens": 2319305.0,
+      "step": 1000
+    },
+    {
+      "epoch": 2.4103802051901027,
+      "eval_entropy": 0.6634890969549672,
+      "eval_loss": 0.6692465543746948,
+      "eval_mean_token_accuracy": 0.8134353673190213,
+      "eval_num_tokens": 2319305.0,
+      "eval_runtime": 87.7711,
+      "eval_samples_per_second": 16.178,
+      "eval_steps_per_second": 2.028,
+      "step": 1000
+    },
+    {
+      "entropy": 0.6169968567788601,
+      "epoch": 2.4586602293301145,
+      "grad_norm": 0.7081006765365601,
+      "learning_rate": 0.0003525243462115402,
+      "loss": 0.5589694023132324,
+      "mean_token_accuracy": 0.8357502184808254,
+      "num_tokens": 2368592.0,
+      "step": 1020
+    },
+    {
+      "epoch": 2.4586602293301145,
+      "eval_entropy": 0.6126393578695447,
+      "eval_loss": 0.6706712245941162,
+      "eval_mean_token_accuracy": 0.8155577500884452,
+      "eval_num_tokens": 2368592.0,
+      "eval_runtime": 88.1407,
+      "eval_samples_per_second": 16.111,
+      "eval_steps_per_second": 2.019,
+      "step": 1020
+    },
+    {
+      "entropy": 0.6072248566895724,
+      "epoch": 2.506940253470127,
+      "grad_norm": 0.6999370455741882,
+      "learning_rate": 0.0003509614855443561,
+      "loss": 0.5500593185424805,
+      "mean_token_accuracy": 0.8337264291942119,
+      "num_tokens": 2412859.0,
+      "step": 1040
+    },
+    {
+      "epoch": 2.506940253470127,
+      "eval_entropy": 0.6417417074187418,
+      "eval_loss": 0.6651163697242737,
+      "eval_mean_token_accuracy": 0.8159113908081912,
+      "eval_num_tokens": 2412859.0,
+      "eval_runtime": 87.4854,
+      "eval_samples_per_second": 16.231,
+      "eval_steps_per_second": 2.035,
+      "step": 1040
+    },
+    {
+      "entropy": 0.6301592070609331,
+      "epoch": 2.5552202776101387,
+      "grad_norm": 0.8768910765647888,
+      "learning_rate": 0.00034935254927581064,
+      "loss": 0.5613903999328613,
+      "mean_token_accuracy": 0.8298896946012974,
+      "num_tokens": 2458201.0,
+      "step": 1060
+    },
+    {
+      "epoch": 2.5552202776101387,
+      "eval_entropy": 0.6569794751285167,
+      "eval_loss": 0.6668341159820557,
+      "eval_mean_token_accuracy": 0.8126791795987761,
+      "eval_num_tokens": 2458201.0,
+      "eval_runtime": 88.0474,
+      "eval_samples_per_second": 16.128,
+      "eval_steps_per_second": 2.022,
+      "step": 1060
+    },
+    {
+      "entropy": 0.6069021724164486,
+      "epoch": 2.603500301750151,
+      "grad_norm": 0.9033508896827698,
+      "learning_rate": 0.0003476979927158357,
+      "loss": 0.5590654373168945,
+      "mean_token_accuracy": 0.831520090252161,
+      "num_tokens": 2505382.0,
+      "step": 1080
+    },
+    {
+      "epoch": 2.603500301750151,
+      "eval_entropy": 0.6636380777600106,
+      "eval_loss": 0.6596238017082214,
+      "eval_mean_token_accuracy": 0.8177202826135614,
+      "eval_num_tokens": 2505382.0,
+      "eval_runtime": 88.08,
+      "eval_samples_per_second": 16.122,
+      "eval_steps_per_second": 2.021,
+      "step": 1080
+    },
+    {
+      "entropy": 0.6138021353632211,
+      "epoch": 2.651780325890163,
+      "grad_norm": 0.7345808148384094,
+      "learning_rate": 0.0003459982840843664,
+      "loss": 0.5586410522460937,
+      "mean_token_accuracy": 0.8295876093208789,
+      "num_tokens": 2552927.0,
+      "step": 1100
+    },
+    {
+      "epoch": 2.651780325890163,
+      "eval_entropy": 0.5845904127600487,
+      "eval_loss": 0.663780689239502,
+      "eval_mean_token_accuracy": 0.8142095107710763,
+      "eval_num_tokens": 2552927.0,
+      "eval_runtime": 88.0083,
+      "eval_samples_per_second": 16.135,
+      "eval_steps_per_second": 2.023,
+      "step": 1100
+    }
+  ],
+  "logging_steps": 20,
+  "max_steps": 4150,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 10,
+  "save_steps": 20,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 1.0779277426032845e+17,
+  "train_batch_size": 4,
+  "trial_name": null,
+  "trial_params": null
+}

overgeneralisation_original_Estonian/Qwen3.5-4B-Base_overgeneralisation_splits_original_features_train_overgeneralisation_splits_original_features_test2/checkpoint-1120/README.md

@@ -0,0 +1,209 @@
+---
+base_model: Qwen/Qwen3.5-4B-Base
+library_name: peft
+pipeline_tag: text-generation
+tags:
+- base_model:adapter:Qwen/Qwen3.5-4B-Base
+- lora
+- sft
+- transformers
+- trl
+---
+
+# Model Card for Model ID
+
+<!-- Provide a quick summary of what the model is/does. -->
+
+
+
+## Model Details
+
+### Model Description
+
+<!-- Provide a longer summary of what this model is. -->
+
+
+
+- **Developed by:** [More Information Needed]
+- **Funded by [optional]:** [More Information Needed]
+- **Shared by [optional]:** [More Information Needed]
+- **Model type:** [More Information Needed]
+- **Language(s) (NLP):** [More Information Needed]
+- **License:** [More Information Needed]
+- **Finetuned from model [optional]:** [More Information Needed]
+
+### Model Sources [optional]
+
+<!-- Provide the basic links for the model. -->
+
+- **Repository:** [More Information Needed]
+- **Paper [optional]:** [More Information Needed]
+- **Demo [optional]:** [More Information Needed]
+
+## Uses
+
+<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
+
+### Direct Use
+
+<!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
+
+[More Information Needed]
+
+### Downstream Use [optional]
+
+<!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->
+
+[More Information Needed]
+
+### Out-of-Scope Use
+
+<!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
+
+[More Information Needed]
+
+## Bias, Risks, and Limitations
+
+<!-- This section is meant to convey both technical and sociotechnical limitations. -->
+
+[More Information Needed]
+
+### Recommendations
+
+<!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->
+
+Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
+
+## How to Get Started with the Model
+
+Use the code below to get started with the model.
+
+[More Information Needed]
+
+## Training Details
+
+### Training Data
+
+<!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
+
+[More Information Needed]
+
+### Training Procedure
+
+<!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
+
+#### Preprocessing [optional]
+
+[More Information Needed]
+
+
+#### Training Hyperparameters
+
+- **Training regime:** [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->
+
+#### Speeds, Sizes, Times [optional]
+
+<!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
+
+[More Information Needed]
+
+## Evaluation
+
+<!-- This section describes the evaluation protocols and provides the results. -->
+
+### Testing Data, Factors & Metrics
+
+#### Testing Data
+
+<!-- This should link to a Dataset Card if possible. -->
+
+[More Information Needed]
+
+#### Factors
+
+<!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->
+
+[More Information Needed]
+
+#### Metrics
+
+<!-- These are the evaluation metrics being used, ideally with a description of why. -->
+
+[More Information Needed]
+
+### Results
+
+[More Information Needed]
+
+#### Summary
+
+
+
+## Model Examination [optional]
+
+<!-- Relevant interpretability work for the model goes here -->
+
+[More Information Needed]
+
+## Environmental Impact
+
+<!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly -->
+
+Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700).
+
+- **Hardware Type:** [More Information Needed]
+- **Hours used:** [More Information Needed]
+- **Cloud Provider:** [More Information Needed]
+- **Compute Region:** [More Information Needed]
+- **Carbon Emitted:** [More Information Needed]
+
+## Technical Specifications [optional]
+
+### Model Architecture and Objective
+
+[More Information Needed]
+
+### Compute Infrastructure
+
+[More Information Needed]
+
+#### Hardware
+
+[More Information Needed]
+
+#### Software
+
+[More Information Needed]
+
+## Citation [optional]
+
+<!-- If there is a paper or blog post introducing the model, the APA and Bibtex information for that should go in this section. -->
+
+**BibTeX:**
+
+[More Information Needed]
+
+**APA:**
+
+[More Information Needed]
+
+## Glossary [optional]
+
+<!-- If relevant, include terms and calculations in this section that can help readers understand the model or model card. -->
+
+[More Information Needed]
+
+## More Information [optional]
+
+[More Information Needed]
+
+## Model Card Authors [optional]
+
+[More Information Needed]
+
+## Model Card Contact
+
+[More Information Needed]
+### Framework versions
+
+- PEFT 0.18.1

overgeneralisation_original_Estonian/Qwen3.5-4B-Base_overgeneralisation_splits_original_features_train_overgeneralisation_splits_original_features_test2/checkpoint-1120/adapter_config.json

@@ -0,0 +1,46 @@
+{
+  "alora_invocation_tokens": null,
+  "alpha_pattern": {},
+  "arrow_config": null,
+  "auto_mapping": null,
+  "base_model_name_or_path": "Qwen/Qwen3.5-4B-Base",
+  "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": 128,
+  "lora_bias": false,
+  "lora_dropout": 0.03828026524568501,
+  "megatron_config": null,
+  "megatron_core": "megatron.core",
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "peft_version": "0.18.1",
+  "qalora_group_size": 16,
+  "r": 64,
+  "rank_pattern": {},
+  "revision": null,
+  "target_modules": [
+    "gate_proj",
+    "v_proj",
+    "up_proj",
+    "down_proj",
+    "o_proj",
+    "q_proj",
+    "k_proj"
+  ],
+  "target_parameters": null,
+  "task_type": "CAUSAL_LM",
+  "trainable_token_indices": null,
+  "use_dora": false,
+  "use_qalora": false,
+  "use_rslora": false
+}

overgeneralisation_original_Estonian/Qwen3.5-4B-Base_overgeneralisation_splits_original_features_train_overgeneralisation_splits_original_features_test2/checkpoint-1120/chat_template.jinja

@@ -0,0 +1,154 @@
+{%- set image_count = namespace(value=0) %}
+{%- set video_count = namespace(value=0) %}
+{%- macro render_content(content, do_vision_count, is_system_content=false) %}
+    {%- if content is string %}
+        {{- content }}
+    {%- elif content is iterable and content is not mapping %}
+        {%- for item in content %}
+            {%- if 'image' in item or 'image_url' in item or item.type == 'image' %}
+                {%- if is_system_content %}
+                    {{- raise_exception('System message cannot contain images.') }}
+                {%- endif %}
+                {%- if do_vision_count %}
+                    {%- set image_count.value = image_count.value + 1 %}
+                {%- endif %}
+                {%- if add_vision_id %}
+                    {{- 'Picture ' ~ image_count.value ~ ': ' }}
+                {%- endif %}
+                {{- '<|vision_start|><|image_pad|><|vision_end|>' }}
+            {%- elif 'video' in item or item.type == 'video' %}
+                {%- if is_system_content %}
+                    {{- raise_exception('System message cannot contain videos.') }}
+                {%- endif %}
+                {%- if do_vision_count %}
+                    {%- set video_count.value = video_count.value + 1 %}
+                {%- endif %}
+                {%- if add_vision_id %}
+                    {{- 'Video ' ~ video_count.value ~ ': ' }}
+                {%- endif %}
+                {{- '<|vision_start|><|video_pad|><|vision_end|>' }}
+            {%- elif 'text' in item %}
+                {{- item.text }}
+            {%- else %}
+                {{- raise_exception('Unexpected item type in content.') }}
+            {%- endif %}
+        {%- endfor %}
+    {%- elif content is none or content is undefined %}
+        {{- '' }}
+    {%- else %}
+        {{- raise_exception('Unexpected content type.') }}
+    {%- endif %}
+{%- endmacro %}
+{%- if not messages %}
+    {{- raise_exception('No messages provided.') }}
+{%- endif %}
+{%- if tools and tools is iterable and tools is not mapping %}
+    {{- '<|im_start|>system\n' }}
+    {{- "# Tools\n\nYou have access to the following functions:\n\n<tools>" }}
+    {%- for tool in tools %}
+        {{- "\n" }}
+        {{- tool | tojson }}
+    {%- endfor %}
+    {{- "\n</tools>" }}
+    {{- '\n\nIf you choose to call a function ONLY reply in the following format with NO suffix:\n\n<tool_call>\n<function=example_function_name>\n<parameter=example_parameter_1>\nvalue_1\n</parameter>\n<parameter=example_parameter_2>\nThis is the value for the second parameter\nthat can span\nmultiple lines\n</parameter>\n</function>\n</tool_call>\n\n<IMPORTANT>\nReminder:\n- Function calls MUST follow the specified format: an inner <function=...></function> block must be nested within <tool_call></tool_call> XML tags\n- Required parameters MUST be specified\n- You may provide optional reasoning for your function call in natural language BEFORE the function call, but NOT after\n- If there is no function call available, answer the question like normal with your current knowledge and do not tell the user about function calls\n</IMPORTANT>' }}
+    {%- if messages[0].role == 'system' %}
+        {%- set content = render_content(messages[0].content, false, true)|trim %}
+        {%- if content %}
+            {{- '\n\n' + content }}
+        {%- endif %}
+    {%- endif %}
+    {{- '<|im_end|>\n' }}
+{%- else %}
+    {%- if messages[0].role == 'system' %}
+        {%- set content = render_content(messages[0].content, false, true)|trim %}
+        {{- '<|im_start|>system\n' + content + '<|im_end|>\n' }}
+    {%- endif %}
+{%- endif %}
+{%- set ns = namespace(multi_step_tool=true, last_query_index=messages|length - 1) %}
+{%- for message in messages[::-1] %}
+    {%- set index = (messages|length - 1) - loop.index0 %}
+    {%- if ns.multi_step_tool and message.role == "user" %}
+        {%- set content = render_content(message.content, false)|trim %}
+        {%- if not(content.startswith('<tool_response>') and content.endswith('</tool_response>')) %}
+            {%- set ns.multi_step_tool = false %}
+            {%- set ns.last_query_index = index %}
+        {%- endif %}
+    {%- endif %}
+{%- endfor %}
+{%- if ns.multi_step_tool %}
+    {{- raise_exception('No user query found in messages.') }}
+{%- endif %}
+{%- for message in messages %}
+    {%- set content = render_content(message.content, true)|trim %}
+    {%- if message.role == "system" %}
+        {%- if not loop.first %}
+            {{- raise_exception('System message must be at the beginning.') }}
+        {%- endif %}
+    {%- elif message.role == "user" %}
+        {{- '<|im_start|>' + message.role + '\n' + content + '<|im_end|>' + '\n' }}
+    {%- elif message.role == "assistant" %}
+        {%- set reasoning_content = '' %}
+        {%- if message.reasoning_content is string %}
+            {%- set reasoning_content = message.reasoning_content %}
+        {%- else %}
+            {%- if '</think>' in content %}
+                {%- set reasoning_content = content.split('</think>')[0].rstrip('\n').split('<think>')[-1].lstrip('\n') %}
+                {%- set content = content.split('</think>')[-1].lstrip('\n') %}
+            {%- endif %}
+        {%- endif %}
+        {%- set reasoning_content = reasoning_content|trim %}
+        {%- if loop.index0 > ns.last_query_index %}
+            {{- '<|im_start|>' + message.role + '\n<think>\n' + reasoning_content + '\n</think>\n\n' + content }}
+        {%- else %}
+            {{- '<|im_start|>' + message.role + '\n' + content }}
+        {%- endif %}
+        {%- if message.tool_calls and message.tool_calls is iterable and message.tool_calls is not mapping %}
+            {%- for tool_call in message.tool_calls %}
+                {%- if tool_call.function is defined %}
+                    {%- set tool_call = tool_call.function %}
+                {%- endif %}
+                {%- if loop.first %}
+                    {%- if content|trim %}
+                        {{- '\n\n<tool_call>\n<function=' + tool_call.name + '>\n' }}
+                    {%- else %}
+                        {{- '<tool_call>\n<function=' + tool_call.name + '>\n' }}
+                    {%- endif %}
+                {%- else %}
+                    {{- '\n<tool_call>\n<function=' + tool_call.name + '>\n' }}
+                {%- endif %}
+                {%- if tool_call.arguments is defined %}
+                    {%- for args_name, args_value in tool_call.arguments|items %}
+                        {{- '<parameter=' + args_name + '>\n' }}
+                        {%- set args_value = args_value | tojson | safe if args_value is mapping or (args_value is sequence and args_value is not string) else args_value | string %}
+                        {{- args_value }}
+                        {{- '\n</parameter>\n' }}
+                    {%- endfor %}
+                {%- endif %}
+                {{- '</function>\n</tool_call>' }}
+            {%- endfor %}
+        {%- endif %}
+        {{- '<|im_end|>\n' }}
+    {%- elif message.role == "tool" %}
+        {%- if loop.previtem and loop.previtem.role != "tool" %}
+            {{- '<|im_start|>user' }}
+        {%- endif %}
+        {{- '\n<tool_response>\n' }}
+        {{- content }}
+        {{- '\n</tool_response>' }}
+        {%- if not loop.last and loop.nextitem.role != "tool" %}
+            {{- '<|im_end|>\n' }}
+        {%- elif loop.last %}
+            {{- '<|im_end|>\n' }}
+        {%- endif %}
+    {%- else %}
+        {{- raise_exception('Unexpected message role.') }}
+    {%- endif %}
+{%- endfor %}
+{%- if add_generation_prompt %}
+    {{- '<|im_start|>assistant\n' }}
+    {%- if enable_thinking is defined and enable_thinking is false %}
+        {{- '<think>\n\n</think>\n\n' }}
+    {%- else %}
+        {{- '<think>\n' }}
+    {%- endif %}
+{%- endif %}

overgeneralisation_original_Estonian/Qwen3.5-4B-Base_overgeneralisation_splits_original_features_train_overgeneralisation_splits_original_features_test2/checkpoint-1120/tokenizer_config.json

@@ -0,0 +1,31 @@
+{
+  "add_prefix_space": false,
+  "audio_bos_token": "<|audio_start|>",
+  "audio_eos_token": "<|audio_end|>",
+  "audio_token": "<|audio_pad|>",
+  "backend": "tokenizers",
+  "bos_token": null,
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<|endoftext|>",
+  "errors": "replace",
+  "image_token": "<|image_pad|>",
+  "is_local": false,
+  "model_max_length": 262144,
+  "model_specific_special_tokens": {
+    "audio_bos_token": "<|audio_start|>",
+    "audio_eos_token": "<|audio_end|>",
+    "audio_token": "<|audio_pad|>",
+    "image_token": "<|image_pad|>",
+    "video_token": "<|video_pad|>",
+    "vision_bos_token": "<|vision_start|>",
+    "vision_eos_token": "<|vision_end|>"
+  },
+  "pad_token": "<|endoftext|>",
+  "pretokenize_regex": "(?i:'s|'t|'re|'ve|'m|'ll|'d)|[^\\r\\n\\p{L}\\p{N}]?[\\p{L}\\p{M}]+|\\p{N}| ?[^\\s\\p{L}\\p{M}\\p{N}]+[\\r\\n]*|\\s*[\\r\\n]+|\\s+(?!\\S)|\\s+",
+  "split_special_tokens": false,
+  "tokenizer_class": "TokenizersBackend",
+  "unk_token": null,
+  "video_token": "<|video_pad|>",
+  "vision_bos_token": "<|vision_start|>",
+  "vision_eos_token": "<|vision_end|>"
+}

overgeneralisation_original_Estonian/Qwen3.5-4B-Base_overgeneralisation_splits_original_features_train_overgeneralisation_splits_original_features_test2/checkpoint-1120/trainer_state.json

@@ -0,0 +1,1210 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 2.700060350030175,
+  "eval_steps": 20,
+  "global_step": 1120,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "entropy": 2.172999547421932,
+      "epoch": 0.04828002414001207,
+      "grad_norm": 2.5071208477020264,
+      "learning_rate": 1.7227585487708594e-05,
+      "loss": 1.9347333908081055,
+      "mean_token_accuracy": 0.6086816020309925,
+      "num_tokens": 47778.0,
+      "step": 20
+    },
+    {
+      "epoch": 0.04828002414001207,
+      "eval_entropy": 1.6909754791956269,
+      "eval_loss": 1.4728385210037231,
+      "eval_mean_token_accuracy": 0.6773593797442619,
+      "eval_num_tokens": 47778.0,
+      "eval_runtime": 90.1265,
+      "eval_samples_per_second": 15.756,
+      "eval_steps_per_second": 1.975,
+      "step": 20
+    },
+    {
+      "entropy": 1.2014626950025558,
+      "epoch": 0.09656004828002414,
+      "grad_norm": 1.5074543952941895,
+      "learning_rate": 3.5361886001086065e-05,
+      "loss": 1.0867146492004394,
+      "mean_token_accuracy": 0.7285927519202232,
+      "num_tokens": 95981.0,
+      "step": 40
+    },
+    {
+      "epoch": 0.09656004828002414,
+      "eval_entropy": 1.0304168352250302,
+      "eval_loss": 0.9211422204971313,
+      "eval_mean_token_accuracy": 0.7613122848312507,
+      "eval_num_tokens": 95981.0,
+      "eval_runtime": 88.6338,
+      "eval_samples_per_second": 16.021,
+      "eval_steps_per_second": 2.008,
+      "step": 40
+    },
+    {
+      "entropy": 0.9454198732972146,
+      "epoch": 0.14484007242003621,
+      "grad_norm": 1.7578108310699463,
+      "learning_rate": 5.349618651446354e-05,
+      "loss": 0.8447297096252442,
+      "mean_token_accuracy": 0.7723424732685089,
+      "num_tokens": 142784.0,
+      "step": 60
+    },
+    {
+      "epoch": 0.14484007242003621,
+      "eval_entropy": 0.8647834474451086,
+      "eval_loss": 0.8240737318992615,
+      "eval_mean_token_accuracy": 0.7821227014064789,
+      "eval_num_tokens": 142784.0,
+      "eval_runtime": 88.0135,
+      "eval_samples_per_second": 16.134,
+      "eval_steps_per_second": 2.022,
+      "step": 60
+    },
+    {
+      "entropy": 0.8963550612330436,
+      "epoch": 0.19312009656004828,
+      "grad_norm": 1.1464931964874268,
+      "learning_rate": 7.1630487027841e-05,
+      "loss": 0.7979836463928223,
+      "mean_token_accuracy": 0.7822489373385906,
+      "num_tokens": 187338.0,
+      "step": 80
+    },
+    {
+      "epoch": 0.19312009656004828,
+      "eval_entropy": 0.8663296133614657,
+      "eval_loss": 0.7752988338470459,
+      "eval_mean_token_accuracy": 0.7885957012015782,
+      "eval_num_tokens": 187338.0,
+      "eval_runtime": 88.284,
+      "eval_samples_per_second": 16.084,
+      "eval_steps_per_second": 2.016,
+      "step": 80
+    },
+    {
+      "entropy": 0.8265103206038475,
+      "epoch": 0.24140012070006034,
+      "grad_norm": 1.0106662511825562,
+      "learning_rate": 8.976478754121848e-05,
+      "loss": 0.7297060012817382,
+      "mean_token_accuracy": 0.7955138415098191,
+      "num_tokens": 236038.0,
+      "step": 100
+    },
+    {
+      "epoch": 0.24140012070006034,
+      "eval_entropy": 0.8122169536151244,
+      "eval_loss": 0.754906177520752,
+      "eval_mean_token_accuracy": 0.7917155255092664,
+      "eval_num_tokens": 236038.0,
+      "eval_runtime": 87.5754,
+      "eval_samples_per_second": 16.215,
+      "eval_steps_per_second": 2.033,
+      "step": 100
+    },
+    {
+      "entropy": 0.8098392844200134,
+      "epoch": 0.28968014484007243,
+      "grad_norm": 1.2238041162490845,
+      "learning_rate": 0.00010789908805459595,
+      "loss": 0.7201489448547364,
+      "mean_token_accuracy": 0.7975016921758652,
+      "num_tokens": 282918.0,
+      "step": 120
+    },
+    {
+      "epoch": 0.28968014484007243,
+      "eval_entropy": 0.8418726462326692,
+      "eval_loss": 0.7393178939819336,
+      "eval_mean_token_accuracy": 0.7936771585700217,
+      "eval_num_tokens": 282918.0,
+      "eval_runtime": 87.1597,
+      "eval_samples_per_second": 16.292,
+      "eval_steps_per_second": 2.042,
+      "step": 120
+    },
+    {
+      "entropy": 0.8080591082572937,
+      "epoch": 0.33796016898008446,
+      "grad_norm": 1.011591911315918,
+      "learning_rate": 0.00012603338856797343,
+      "loss": 0.7147697925567627,
+      "mean_token_accuracy": 0.7947175532579422,
+      "num_tokens": 331305.0,
+      "step": 140
+    },
+    {
+      "epoch": 0.33796016898008446,
+      "eval_entropy": 0.7894941431083037,
+      "eval_loss": 0.720219075679779,
+      "eval_mean_token_accuracy": 0.801899236239744,
+      "eval_num_tokens": 331305.0,
+      "eval_runtime": 87.1493,
+      "eval_samples_per_second": 16.294,
+      "eval_steps_per_second": 2.042,
+      "step": 140
+    },
+    {
+      "entropy": 0.7957205109298229,
+      "epoch": 0.38624019312009655,
+      "grad_norm": 0.8481830954551697,
+      "learning_rate": 0.00014416768908135088,
+      "loss": 0.7050027847290039,
+      "mean_token_accuracy": 0.7995368830859662,
+      "num_tokens": 375876.0,
+      "step": 160
+    },
+    {
+      "epoch": 0.38624019312009655,
+      "eval_entropy": 0.7746140591883928,
+      "eval_loss": 0.7175475358963013,
+      "eval_mean_token_accuracy": 0.7994848955213354,
+      "eval_num_tokens": 375876.0,
+      "eval_runtime": 86.655,
+      "eval_samples_per_second": 16.387,
+      "eval_steps_per_second": 2.054,
+      "step": 160
+    },
+    {
+      "entropy": 0.779565304517746,
+      "epoch": 0.43452021726010864,
+      "grad_norm": 1.032624363899231,
+      "learning_rate": 0.00016230198959472835,
+      "loss": 0.6978332042694092,
+      "mean_token_accuracy": 0.8015618294477462,
+      "num_tokens": 423913.0,
+      "step": 180
+    },
+    {
+      "epoch": 0.43452021726010864,
+      "eval_entropy": 0.7455583871080634,
+      "eval_loss": 0.7118850350379944,
+      "eval_mean_token_accuracy": 0.7999682195400923,
+      "eval_num_tokens": 423913.0,
+      "eval_runtime": 86.7842,
+      "eval_samples_per_second": 16.362,
+      "eval_steps_per_second": 2.051,
+      "step": 180
+    },
+    {
+      "entropy": 0.7810172818601131,
+      "epoch": 0.4828002414001207,
+      "grad_norm": 0.9276663064956665,
+      "learning_rate": 0.00018043629010810582,
+      "loss": 0.6937861442565918,
+      "mean_token_accuracy": 0.802893053740263,
+      "num_tokens": 472395.0,
+      "step": 200
+    },
+    {
+      "epoch": 0.4828002414001207,
+      "eval_entropy": 0.7650193290764027,
+      "eval_loss": 0.7020567059516907,
+      "eval_mean_token_accuracy": 0.8045302153973097,
+      "eval_num_tokens": 472395.0,
+      "eval_runtime": 88.326,
+      "eval_samples_per_second": 16.077,
+      "eval_steps_per_second": 2.015,
+      "step": 200
+    },
+    {
+      "entropy": 0.7862060204148292,
+      "epoch": 0.5310802655401328,
+      "grad_norm": 0.8195295929908752,
+      "learning_rate": 0.0001985705906214833,
+      "loss": 0.6878782272338867,
+      "mean_token_accuracy": 0.8027168907225132,
+      "num_tokens": 517049.0,
+      "step": 220
+    },
+    {
+      "epoch": 0.5310802655401328,
+      "eval_entropy": 0.7656565583154057,
+      "eval_loss": 0.6973932981491089,
+      "eval_mean_token_accuracy": 0.8047967278555538,
+      "eval_num_tokens": 517049.0,
+      "eval_runtime": 87.7698,
+      "eval_samples_per_second": 16.179,
+      "eval_steps_per_second": 2.028,
+      "step": 220
+    },
+    {
+      "entropy": 0.7611534461379051,
+      "epoch": 0.5793602896801449,
+      "grad_norm": 0.7499712109565735,
+      "learning_rate": 0.00021670489113486077,
+      "loss": 0.681937837600708,
+      "mean_token_accuracy": 0.8041978515684605,
+      "num_tokens": 562836.0,
+      "step": 240
+    },
+    {
+      "epoch": 0.5793602896801449,
+      "eval_entropy": 0.7599442380197933,
+      "eval_loss": 0.7002046704292297,
+      "eval_mean_token_accuracy": 0.8051861517884759,
+      "eval_num_tokens": 562836.0,
+      "eval_runtime": 87.9239,
+      "eval_samples_per_second": 16.15,
+      "eval_steps_per_second": 2.024,
+      "step": 240
+    },
+    {
+      "entropy": 0.7800783462822437,
+      "epoch": 0.627640313820157,
+      "grad_norm": 0.8645269274711609,
+      "learning_rate": 0.00023483919164823825,
+      "loss": 0.6954500675201416,
+      "mean_token_accuracy": 0.8031484372913837,
+      "num_tokens": 606612.0,
+      "step": 260
+    },
+    {
+      "epoch": 0.627640313820157,
+      "eval_entropy": 0.7498679090751691,
+      "eval_loss": 0.7046149373054504,
+      "eval_mean_token_accuracy": 0.802388215667746,
+      "eval_num_tokens": 606612.0,
+      "eval_runtime": 88.2468,
+      "eval_samples_per_second": 16.091,
+      "eval_steps_per_second": 2.017,
+      "step": 260
+    },
+    {
+      "entropy": 0.7586533665657044,
+      "epoch": 0.6759203379601689,
+      "grad_norm": 0.741596519947052,
+      "learning_rate": 0.0002529734921616157,
+      "loss": 0.690476369857788,
+      "mean_token_accuracy": 0.8046923100948333,
+      "num_tokens": 655920.0,
+      "step": 280
+    },
+    {
+      "epoch": 0.6759203379601689,
+      "eval_entropy": 0.7714274066887544,
+      "eval_loss": 0.6992160677909851,
+      "eval_mean_token_accuracy": 0.8042646482419432,
+      "eval_num_tokens": 655920.0,
+      "eval_runtime": 88.1754,
+      "eval_samples_per_second": 16.104,
+      "eval_steps_per_second": 2.019,
+      "step": 280
+    },
+    {
+      "entropy": 0.7585848286747933,
+      "epoch": 0.724200362100181,
+      "grad_norm": 1.0778124332427979,
+      "learning_rate": 0.00027110779267499317,
+      "loss": 0.6790409564971924,
+      "mean_token_accuracy": 0.8060351841151714,
+      "num_tokens": 702638.0,
+      "step": 300
+    },
+    {
+      "epoch": 0.724200362100181,
+      "eval_entropy": 0.7380402811457601,
+      "eval_loss": 0.7054820656776428,
+      "eval_mean_token_accuracy": 0.8003534738267406,
+      "eval_num_tokens": 702638.0,
+      "eval_runtime": 88.238,
+      "eval_samples_per_second": 16.093,
+      "eval_steps_per_second": 2.017,
+      "step": 300
+    },
+    {
+      "entropy": 0.7696981988847256,
+      "epoch": 0.7724803862401931,
+      "grad_norm": 0.8408658504486084,
+      "learning_rate": 0.00028924209318837064,
+      "loss": 0.7070387363433838,
+      "mean_token_accuracy": 0.8035947173833847,
+      "num_tokens": 749377.0,
+      "step": 320
+    },
+    {
+      "epoch": 0.7724803862401931,
+      "eval_entropy": 0.7897409023193831,
+      "eval_loss": 0.7015668749809265,
+      "eval_mean_token_accuracy": 0.805306687113944,
+      "eval_num_tokens": 749377.0,
+      "eval_runtime": 87.9395,
+      "eval_samples_per_second": 16.147,
+      "eval_steps_per_second": 2.024,
+      "step": 320
+    },
+    {
+      "entropy": 0.762337189912796,
+      "epoch": 0.8207604103802052,
+      "grad_norm": 0.9843310117721558,
+      "learning_rate": 0.0003073763937017481,
+      "loss": 0.7021088123321533,
+      "mean_token_accuracy": 0.8029616877436638,
+      "num_tokens": 798874.0,
+      "step": 340
+    },
+    {
+      "epoch": 0.8207604103802052,
+      "eval_entropy": 0.7210077121016685,
+      "eval_loss": 0.7082746624946594,
+      "eval_mean_token_accuracy": 0.8049283824609906,
+      "eval_num_tokens": 798874.0,
+      "eval_runtime": 88.1706,
+      "eval_samples_per_second": 16.105,
+      "eval_steps_per_second": 2.019,
+      "step": 340
+    },
+    {
+      "entropy": 0.7879025422036647,
+      "epoch": 0.8690404345202173,
+      "grad_norm": 0.9713814854621887,
+      "learning_rate": 0.0003255106942151256,
+      "loss": 0.714624547958374,
+      "mean_token_accuracy": 0.79987031519413,
+      "num_tokens": 840825.0,
+      "step": 360
+    },
+    {
+      "epoch": 0.8690404345202173,
+      "eval_entropy": 0.7866930489459735,
+      "eval_loss": 0.7145519852638245,
+      "eval_mean_token_accuracy": 0.8015657988157165,
+      "eval_num_tokens": 840825.0,
+      "eval_runtime": 88.3746,
+      "eval_samples_per_second": 16.068,
+      "eval_steps_per_second": 2.014,
+      "step": 360
+    },
+    {
+      "entropy": 0.7844726584851742,
+      "epoch": 0.9173204586602294,
+      "grad_norm": 1.2028311491012573,
+      "learning_rate": 0.00034364499472850306,
+      "loss": 0.7040786266326904,
+      "mean_token_accuracy": 0.7981885217130185,
+      "num_tokens": 882885.0,
+      "step": 380
+    },
+    {
+      "epoch": 0.9173204586602294,
+      "eval_entropy": 0.7767668339643585,
+      "eval_loss": 0.7073574066162109,
+      "eval_mean_token_accuracy": 0.8040851833445303,
+      "eval_num_tokens": 882885.0,
+      "eval_runtime": 88.3828,
+      "eval_samples_per_second": 16.066,
+      "eval_steps_per_second": 2.014,
+      "step": 380
+    },
+    {
+      "entropy": 0.7699790760874748,
+      "epoch": 0.9656004828002414,
+      "grad_norm": 0.9397398233413696,
+      "learning_rate": 0.00036177929524188054,
+      "loss": 0.7040313720703125,
+      "mean_token_accuracy": 0.8034727744758129,
+      "num_tokens": 927456.0,
+      "step": 400
+    },
+    {
+      "epoch": 0.9656004828002414,
+      "eval_entropy": 0.7749103545472863,
+      "eval_loss": 0.7093513011932373,
+      "eval_mean_token_accuracy": 0.8019482901926791,
+      "eval_num_tokens": 927456.0,
+      "eval_runtime": 86.2112,
+      "eval_samples_per_second": 16.471,
+      "eval_steps_per_second": 2.065,
+      "step": 400
+    },
+    {
+      "entropy": 0.7672389172888422,
+      "epoch": 1.012070006035003,
+      "grad_norm": 0.99210524559021,
+      "learning_rate": 0.00037628567078152296,
+      "loss": 0.6913118362426758,
+      "mean_token_accuracy": 0.8042210944287189,
+      "num_tokens": 972662.0,
+      "step": 420
+    },
+    {
+      "epoch": 1.012070006035003,
+      "eval_entropy": 0.7294771229283193,
+      "eval_loss": 0.7171520590782166,
+      "eval_mean_token_accuracy": 0.8015128585059991,
+      "eval_num_tokens": 972662.0,
+      "eval_runtime": 88.037,
+      "eval_samples_per_second": 16.13,
+      "eval_steps_per_second": 2.022,
+      "step": 420
+    },
+    {
+      "entropy": 0.7143943183124065,
+      "epoch": 1.060350030175015,
+      "grad_norm": 1.087638258934021,
+      "learning_rate": 0.0003762484015601069,
+      "loss": 0.642704439163208,
+      "mean_token_accuracy": 0.8124966286122799,
+      "num_tokens": 1022525.0,
+      "step": 440
+    },
+    {
+      "epoch": 1.060350030175015,
+      "eval_entropy": 0.743511886744017,
+      "eval_loss": 0.7116662859916687,
+      "eval_mean_token_accuracy": 0.8037210672758939,
+      "eval_num_tokens": 1022525.0,
+      "eval_runtime": 87.9931,
+      "eval_samples_per_second": 16.138,
+      "eval_steps_per_second": 2.023,
+      "step": 440
+    },
+    {
+      "entropy": 0.7237346574664116,
+      "epoch": 1.1086300543150271,
+      "grad_norm": 0.9365553259849548,
+      "learning_rate": 0.0003761579008392123,
+      "loss": 0.652644681930542,
+      "mean_token_accuracy": 0.8107496216893196,
+      "num_tokens": 1069194.0,
+      "step": 460
+    },
+    {
+      "epoch": 1.1086300543150271,
+      "eval_entropy": 0.7372823839776972,
+      "eval_loss": 0.7238303422927856,
+      "eval_mean_token_accuracy": 0.8026230124275336,
+      "eval_num_tokens": 1069194.0,
+      "eval_runtime": 87.6778,
+      "eval_samples_per_second": 16.196,
+      "eval_steps_per_second": 2.03,
+      "step": 460
+    },
+    {
+      "entropy": 0.7202137842774391,
+      "epoch": 1.1569100784550392,
+      "grad_norm": 0.7998443841934204,
+      "learning_rate": 0.0003760141942294726,
+      "loss": 0.6485219478607178,
+      "mean_token_accuracy": 0.8092714451253414,
+      "num_tokens": 1117570.0,
+      "step": 480
+    },
+    {
+      "epoch": 1.1569100784550392,
+      "eval_entropy": 0.71547647942318,
+      "eval_loss": 0.7124277949333191,
+      "eval_mean_token_accuracy": 0.8013174219077892,
+      "eval_num_tokens": 1117570.0,
+      "eval_runtime": 87.4387,
+      "eval_samples_per_second": 16.24,
+      "eval_steps_per_second": 2.036,
+      "step": 480
+    },
+    {
+      "entropy": 0.7270819500088692,
+      "epoch": 1.2051901025950513,
+      "grad_norm": 0.9470787644386292,
+      "learning_rate": 0.00037581732239815854,
+      "loss": 0.66352219581604,
+      "mean_token_accuracy": 0.8102593503892421,
+      "num_tokens": 1161717.0,
+      "step": 500
+    },
+    {
+      "epoch": 1.2051901025950513,
+      "eval_entropy": 0.7791565897759427,
+      "eval_loss": 0.7179726958274841,
+      "eval_mean_token_accuracy": 0.7948898328154275,
+      "eval_num_tokens": 1161717.0,
+      "eval_runtime": 87.9927,
+      "eval_samples_per_second": 16.138,
+      "eval_steps_per_second": 2.023,
+      "step": 500
+    },
+    {
+      "entropy": 0.7320936664938926,
+      "epoch": 1.2534701267350634,
+      "grad_norm": 1.0220379829406738,
+      "learning_rate": 0.0003755673410576695,
+      "loss": 0.6581549644470215,
+      "mean_token_accuracy": 0.8098709337413311,
+      "num_tokens": 1209390.0,
+      "step": 520
+    },
+    {
+      "epoch": 1.2534701267350634,
+      "eval_entropy": 0.7710678225822662,
+      "eval_loss": 0.708797812461853,
+      "eval_mean_token_accuracy": 0.8039572778042783,
+      "eval_num_tokens": 1209390.0,
+      "eval_runtime": 88.1193,
+      "eval_samples_per_second": 16.115,
+      "eval_steps_per_second": 2.02,
+      "step": 520
+    },
+    {
+      "entropy": 0.7281674664467573,
+      "epoch": 1.3017501508750755,
+      "grad_norm": 0.8775188326835632,
+      "learning_rate": 0.000375264320949768,
+      "loss": 0.6631278991699219,
+      "mean_token_accuracy": 0.8101261422038079,
+      "num_tokens": 1256611.0,
+      "step": 540
+    },
+    {
+      "epoch": 1.3017501508750755,
+      "eval_entropy": 0.7199799800187014,
+      "eval_loss": 0.6981531381607056,
+      "eval_mean_token_accuracy": 0.8071265254127845,
+      "eval_num_tokens": 1256611.0,
+      "eval_runtime": 88.165,
+      "eval_samples_per_second": 16.106,
+      "eval_steps_per_second": 2.019,
+      "step": 540
+    },
+    {
+      "entropy": 0.722860773652792,
+      "epoch": 1.3500301750150876,
+      "grad_norm": 0.8949288129806519,
+      "learning_rate": 0.00037490834782556,
+      "loss": 0.6639598846435547,
+      "mean_token_accuracy": 0.8105649061501026,
+      "num_tokens": 1302325.0,
+      "step": 560
+    },
+    {
+      "epoch": 1.3500301750150876,
+      "eval_entropy": 0.7134760032209118,
+      "eval_loss": 0.7029948234558105,
+      "eval_mean_token_accuracy": 0.8062985880991046,
+      "eval_num_tokens": 1302325.0,
+      "eval_runtime": 88.1286,
+      "eval_samples_per_second": 16.113,
+      "eval_steps_per_second": 2.02,
+      "step": 560
+    },
+    {
+      "entropy": 0.7299126699566841,
+      "epoch": 1.3983101991550995,
+      "grad_norm": 1.0323039293289185,
+      "learning_rate": 0.0003744995224212291,
+      "loss": 0.6535763263702392,
+      "mean_token_accuracy": 0.8120224848389626,
+      "num_tokens": 1343844.0,
+      "step": 580
+    },
+    {
+      "epoch": 1.3983101991550995,
+      "eval_entropy": 0.7169701182440426,
+      "eval_loss": 0.7067857980728149,
+      "eval_mean_token_accuracy": 0.8040432936689826,
+      "eval_num_tokens": 1343844.0,
+      "eval_runtime": 88.2252,
+      "eval_samples_per_second": 16.095,
+      "eval_steps_per_second": 2.018,
+      "step": 580
+    },
+    {
+      "entropy": 0.7281742256134749,
+      "epoch": 1.4465902232951118,
+      "grad_norm": 1.3370424509048462,
+      "learning_rate": 0.00037403796042952863,
+      "loss": 0.6637272357940673,
+      "mean_token_accuracy": 0.8118261776864528,
+      "num_tokens": 1389471.0,
+      "step": 600
+    },
+    {
+      "epoch": 1.4465902232951118,
+      "eval_entropy": 0.7378843212395572,
+      "eval_loss": 0.7038390040397644,
+      "eval_mean_token_accuracy": 0.8059229371922739,
+      "eval_num_tokens": 1389471.0,
+      "eval_runtime": 87.9301,
+      "eval_samples_per_second": 16.149,
+      "eval_steps_per_second": 2.024,
+      "step": 600
+    },
+    {
+      "entropy": 0.7388614989817143,
+      "epoch": 1.4948702474351236,
+      "grad_norm": 0.8464171290397644,
+      "learning_rate": 0.00037352379246704257,
+      "loss": 0.6604740142822265,
+      "mean_token_accuracy": 0.8102974124252796,
+      "num_tokens": 1435975.0,
+      "step": 620
+    },
+    {
+      "epoch": 1.4948702474351236,
+      "eval_entropy": 0.7126847046814607,
+      "eval_loss": 0.6950703263282776,
+      "eval_mean_token_accuracy": 0.8085885392815879,
+      "eval_num_tokens": 1435975.0,
+      "eval_runtime": 88.2732,
+      "eval_samples_per_second": 16.086,
+      "eval_steps_per_second": 2.016,
+      "step": 620
+    },
+    {
+      "entropy": 0.7258916139602661,
+      "epoch": 1.5431502715751357,
+      "grad_norm": 1.1780085563659668,
+      "learning_rate": 0.0003729571640372223,
+      "loss": 0.6529891014099121,
+      "mean_token_accuracy": 0.8085981778800487,
+      "num_tokens": 1482035.0,
+      "step": 640
+    },
+    {
+      "epoch": 1.5431502715751357,
+      "eval_entropy": 0.7096509638797032,
+      "eval_loss": 0.690051257610321,
+      "eval_mean_token_accuracy": 0.8100520241796301,
+      "eval_num_tokens": 1482035.0,
+      "eval_runtime": 88.3153,
+      "eval_samples_per_second": 16.079,
+      "eval_steps_per_second": 2.016,
+      "step": 640
+    },
+    {
+      "entropy": 0.7151173129677773,
+      "epoch": 1.5914302957151478,
+      "grad_norm": 1.0357667207717896,
+      "learning_rate": 0.0003723382354892108,
+      "loss": 0.6451474189758301,
+      "mean_token_accuracy": 0.8136902332305909,
+      "num_tokens": 1528119.0,
+      "step": 660
+    },
+    {
+      "epoch": 1.5914302957151478,
+      "eval_entropy": 0.7060252652409371,
+      "eval_loss": 0.6899483799934387,
+      "eval_mean_token_accuracy": 0.810834194502134,
+      "eval_num_tokens": 1528119.0,
+      "eval_runtime": 87.8848,
+      "eval_samples_per_second": 16.158,
+      "eval_steps_per_second": 2.025,
+      "step": 660
+    },
+    {
+      "entropy": 0.7141377851366997,
+      "epoch": 1.63971031985516,
+      "grad_norm": 0.8391667008399963,
+      "learning_rate": 0.000371667181972466,
+      "loss": 0.6481579780578614,
+      "mean_token_accuracy": 0.8159015104174614,
+      "num_tokens": 1573661.0,
+      "step": 680
+    },
+    {
+      "epoch": 1.63971031985516,
+      "eval_entropy": 0.7598993319473909,
+      "eval_loss": 0.681204080581665,
+      "eval_mean_token_accuracy": 0.8100194586126992,
+      "eval_num_tokens": 1573661.0,
+      "eval_runtime": 88.2415,
+      "eval_samples_per_second": 16.092,
+      "eval_steps_per_second": 2.017,
+      "step": 680
+    },
+    {
+      "entropy": 0.7235980287194252,
+      "epoch": 1.687990343995172,
+      "grad_norm": 0.8197196125984192,
+      "learning_rate": 0.00037094419338719537,
+      "loss": 0.6582849025726318,
+      "mean_token_accuracy": 0.8114834539592266,
+      "num_tokens": 1621714.0,
+      "step": 700
+    },
+    {
+      "epoch": 1.687990343995172,
+      "eval_entropy": 0.7166854568411795,
+      "eval_loss": 0.679177463054657,
+      "eval_mean_token_accuracy": 0.8119039877077167,
+      "eval_num_tokens": 1621714.0,
+      "eval_runtime": 88.0957,
+      "eval_samples_per_second": 16.119,
+      "eval_steps_per_second": 2.021,
+      "step": 700
+    },
+    {
+      "entropy": 0.7278190158307553,
+      "epoch": 1.736270368135184,
+      "grad_norm": 0.7820602655410767,
+      "learning_rate": 0.0003701694743306164,
+      "loss": 0.6633153438568116,
+      "mean_token_accuracy": 0.8123249113559723,
+      "num_tokens": 1672118.0,
+      "step": 720
+    },
+    {
+      "epoch": 1.736270368135184,
+      "eval_entropy": 0.6980850314826108,
+      "eval_loss": 0.6775653958320618,
+      "eval_mean_token_accuracy": 0.8116872906684875,
+      "eval_num_tokens": 1672118.0,
+      "eval_runtime": 87.9795,
+      "eval_samples_per_second": 16.14,
+      "eval_steps_per_second": 2.023,
+      "step": 720
+    },
+    {
+      "entropy": 0.7149631556123495,
+      "epoch": 1.7845503922751962,
+      "grad_norm": 1.0139760971069336,
+      "learning_rate": 0.00036934324403905795,
+      "loss": 0.6468976020812989,
+      "mean_token_accuracy": 0.8117561548948288,
+      "num_tokens": 1720465.0,
+      "step": 740
+    },
+    {
+      "epoch": 1.7845503922751962,
+      "eval_entropy": 0.69009849094273,
+      "eval_loss": 0.6731093525886536,
+      "eval_mean_token_accuracy": 0.81281964196248,
+      "eval_num_tokens": 1720465.0,
+      "eval_runtime": 87.7412,
+      "eval_samples_per_second": 16.184,
+      "eval_steps_per_second": 2.029,
+      "step": 740
+    },
+    {
+      "entropy": 0.7226049453020096,
+      "epoch": 1.832830416415208,
+      "grad_norm": 1.5040546655654907,
+      "learning_rate": 0.0003684657363259193,
+      "loss": 0.6385328769683838,
+      "mean_token_accuracy": 0.8132639616727829,
+      "num_tokens": 1764440.0,
+      "step": 760
+    },
+    {
+      "epoch": 1.832830416415208,
+      "eval_entropy": 0.7125169618076153,
+      "eval_loss": 0.6864680647850037,
+      "eval_mean_token_accuracy": 0.8049418089095126,
+      "eval_num_tokens": 1764440.0,
+      "eval_runtime": 87.9429,
+      "eval_samples_per_second": 16.147,
+      "eval_steps_per_second": 2.024,
+      "step": 760
+    },
+    {
+      "entropy": 0.707821810990572,
+      "epoch": 1.8811104405552204,
+      "grad_norm": 0.8604223132133484,
+      "learning_rate": 0.00036753719951550327,
+      "loss": 0.6397049427032471,
+      "mean_token_accuracy": 0.8137877315282822,
+      "num_tokens": 1810828.0,
+      "step": 780
+    },
+    {
+      "epoch": 1.8811104405552204,
+      "eval_entropy": 0.6492102551326323,
+      "eval_loss": 0.6743726134300232,
+      "eval_mean_token_accuracy": 0.8130356572317273,
+      "eval_num_tokens": 1810828.0,
+      "eval_runtime": 87.8684,
+      "eval_samples_per_second": 16.161,
+      "eval_steps_per_second": 2.026,
+      "step": 780
+    },
+    {
+      "entropy": 0.6800355084240437,
+      "epoch": 1.9293904646952322,
+      "grad_norm": 0.7270982265472412,
+      "learning_rate": 0.00036655789637274377,
+      "loss": 0.6267755031585693,
+      "mean_token_accuracy": 0.8180945307016373,
+      "num_tokens": 1857600.0,
+      "step": 800
+    },
+    {
+      "epoch": 1.9293904646952322,
+      "eval_entropy": 0.6727538651294922,
+      "eval_loss": 0.6696028709411621,
+      "eval_mean_token_accuracy": 0.8116443090224534,
+      "eval_num_tokens": 1857600.0,
+      "eval_runtime": 88.2449,
+      "eval_samples_per_second": 16.092,
+      "eval_steps_per_second": 2.017,
+      "step": 800
+    },
+    {
+      "entropy": 0.6746691606938839,
+      "epoch": 1.9776704888352445,
+      "grad_norm": 0.9771540760993958,
+      "learning_rate": 0.0003655281040288461,
+      "loss": 0.6326992988586426,
+      "mean_token_accuracy": 0.8179976396262646,
+      "num_tokens": 1901998.0,
+      "step": 820
+    },
+    {
+      "epoch": 1.9776704888352445,
+      "eval_entropy": 0.6636319716324967,
+      "eval_loss": 0.6714363098144531,
+      "eval_mean_token_accuracy": 0.8126635943235976,
+      "eval_num_tokens": 1901998.0,
+      "eval_runtime": 87.7666,
+      "eval_samples_per_second": 16.179,
+      "eval_steps_per_second": 2.028,
+      "step": 820
+    },
+    {
+      "entropy": 0.6319125778489298,
+      "epoch": 2.024140012070006,
+      "grad_norm": 0.8380444645881653,
+      "learning_rate": 0.0003644481139028622,
+      "loss": 0.5780903816223144,
+      "mean_token_accuracy": 0.8250188401767186,
+      "num_tokens": 1947553.0,
+      "step": 840
+    },
+    {
+      "epoch": 2.024140012070006,
+      "eval_entropy": 0.6627151989535,
+      "eval_loss": 0.6774935722351074,
+      "eval_mean_token_accuracy": 0.8140510819601209,
+      "eval_num_tokens": 1947553.0,
+      "eval_runtime": 87.8287,
+      "eval_samples_per_second": 16.168,
+      "eval_steps_per_second": 2.027,
+      "step": 840
+    },
+    {
+      "entropy": 0.6059147048741579,
+      "epoch": 2.0724200362100182,
+      "grad_norm": 0.9745152592658997,
+      "learning_rate": 0.0003633182316192229,
+      "loss": 0.546702527999878,
+      "mean_token_accuracy": 0.8358186542987823,
+      "num_tokens": 1995234.0,
+      "step": 860
+    },
+    {
+      "epoch": 2.0724200362100182,
+      "eval_entropy": 0.6620710384979677,
+      "eval_loss": 0.6769291162490845,
+      "eval_mean_token_accuracy": 0.8135082835561773,
+      "eval_num_tokens": 1995234.0,
+      "eval_runtime": 88.0008,
+      "eval_samples_per_second": 16.136,
+      "eval_steps_per_second": 2.023,
+      "step": 860
+    },
+    {
+      "entropy": 0.6076117843389511,
+      "epoch": 2.12070006035003,
+      "grad_norm": 0.6914380192756653,
+      "learning_rate": 0.0003621387769212491,
+      "loss": 0.5324535846710206,
+      "mean_token_accuracy": 0.8366568259894848,
+      "num_tokens": 2040639.0,
+      "step": 880
+    },
+    {
+      "epoch": 2.12070006035003,
+      "eval_entropy": 0.6592224213514435,
+      "eval_loss": 0.670383632183075,
+      "eval_mean_token_accuracy": 0.8162150088320957,
+      "eval_num_tokens": 2040639.0,
+      "eval_runtime": 88.324,
+      "eval_samples_per_second": 16.077,
+      "eval_steps_per_second": 2.015,
+      "step": 880
+    },
+    {
+      "entropy": 0.6158072650432587,
+      "epoch": 2.1689800844900424,
+      "grad_norm": 0.9649198055267334,
+      "learning_rate": 0.0003609100835806688,
+      "loss": 0.5564568996429443,
+      "mean_token_accuracy": 0.8329654954373836,
+      "num_tokens": 2090604.0,
+      "step": 900
+    },
+    {
+      "epoch": 2.1689800844900424,
+      "eval_entropy": 0.608588819758276,
+      "eval_loss": 0.6745102405548096,
+      "eval_mean_token_accuracy": 0.8155080542135774,
+      "eval_num_tokens": 2090604.0,
+      "eval_runtime": 87.52,
+      "eval_samples_per_second": 16.225,
+      "eval_steps_per_second": 2.034,
+      "step": 900
+    },
+    {
+      "entropy": 0.5938734326511621,
+      "epoch": 2.2172601086300543,
+      "grad_norm": 0.6915557384490967,
+      "learning_rate": 0.0003596324993031632,
+      "loss": 0.5351875782012939,
+      "mean_token_accuracy": 0.8385631121695042,
+      "num_tokens": 2142246.0,
+      "step": 920
+    },
+    {
+      "epoch": 2.2172601086300543,
+      "eval_entropy": 0.6251631205001574,
+      "eval_loss": 0.6664518117904663,
+      "eval_mean_token_accuracy": 0.81711940417129,
+      "eval_num_tokens": 2142246.0,
+      "eval_runtime": 87.1792,
+      "eval_samples_per_second": 16.288,
+      "eval_steps_per_second": 2.042,
+      "step": 920
+    },
+    {
+      "entropy": 0.623173613846302,
+      "epoch": 2.2655401327700666,
+      "grad_norm": 0.8366610407829285,
+      "learning_rate": 0.00035830638562997063,
+      "loss": 0.5488213539123535,
+      "mean_token_accuracy": 0.8321483485400677,
+      "num_tokens": 2183144.0,
+      "step": 940
+    },
+    {
+      "epoch": 2.2655401327700666,
+      "eval_entropy": 0.6237085649136747,
+      "eval_loss": 0.68075031042099,
+      "eval_mean_token_accuracy": 0.814671738429016,
+      "eval_num_tokens": 2183144.0,
+      "eval_runtime": 88.0652,
+      "eval_samples_per_second": 16.124,
+      "eval_steps_per_second": 2.021,
+      "step": 940
+    },
+    {
+      "entropy": 0.6236911740154027,
+      "epoch": 2.3138201569100785,
+      "grad_norm": 0.7812452912330627,
+      "learning_rate": 0.00035693211783557416,
+      "loss": 0.5696200847625732,
+      "mean_token_accuracy": 0.82991396561265,
+      "num_tokens": 2228098.0,
+      "step": 960
+    },
+    {
+      "epoch": 2.3138201569100785,
+      "eval_entropy": 0.6327033426319615,
+      "eval_loss": 0.6742984652519226,
+      "eval_mean_token_accuracy": 0.8148086091105857,
+      "eval_num_tokens": 2228098.0,
+      "eval_runtime": 88.1711,
+      "eval_samples_per_second": 16.105,
+      "eval_steps_per_second": 2.019,
+      "step": 960
+    },
+    {
+      "entropy": 0.6218426413834095,
+      "epoch": 2.3621001810500903,
+      "grad_norm": 0.8719790577888489,
+      "learning_rate": 0.0003555100848215035,
+      "loss": 0.5582651615142822,
+      "mean_token_accuracy": 0.8323961742222309,
+      "num_tokens": 2273062.0,
+      "step": 980
+    },
+    {
+      "epoch": 2.3621001810500903,
+      "eval_entropy": 0.6338012375858393,
+      "eval_loss": 0.6685822606086731,
+      "eval_mean_token_accuracy": 0.8152180300669724,
+      "eval_num_tokens": 2273062.0,
+      "eval_runtime": 88.101,
+      "eval_samples_per_second": 16.118,
+      "eval_steps_per_second": 2.02,
+      "step": 980
+    },
+    {
+      "entropy": 0.6050681818276644,
+      "epoch": 2.4103802051901027,
+      "grad_norm": 0.9151566028594971,
+      "learning_rate": 0.00035404068900628076,
+      "loss": 0.559452486038208,
+      "mean_token_accuracy": 0.8310242861509323,
+      "num_tokens": 2319305.0,
+      "step": 1000
+    },
+    {
+      "epoch": 2.4103802051901027,
+      "eval_entropy": 0.6634890969549672,
+      "eval_loss": 0.6692465543746948,
+      "eval_mean_token_accuracy": 0.8134353673190213,
+      "eval_num_tokens": 2319305.0,
+      "eval_runtime": 87.7711,
+      "eval_samples_per_second": 16.178,
+      "eval_steps_per_second": 2.028,
+      "step": 1000
+    },
+    {
+      "entropy": 0.6169968567788601,
+      "epoch": 2.4586602293301145,
+      "grad_norm": 0.7081006765365601,
+      "learning_rate": 0.0003525243462115402,
+      "loss": 0.5589694023132324,
+      "mean_token_accuracy": 0.8357502184808254,
+      "num_tokens": 2368592.0,
+      "step": 1020
+    },
+    {
+      "epoch": 2.4586602293301145,
+      "eval_entropy": 0.6126393578695447,
+      "eval_loss": 0.6706712245941162,
+      "eval_mean_token_accuracy": 0.8155577500884452,
+      "eval_num_tokens": 2368592.0,
+      "eval_runtime": 88.1407,
+      "eval_samples_per_second": 16.111,
+      "eval_steps_per_second": 2.019,
+      "step": 1020
+    },
+    {
+      "entropy": 0.6072248566895724,
+      "epoch": 2.506940253470127,
+      "grad_norm": 0.6999370455741882,
+      "learning_rate": 0.0003509614855443561,
+      "loss": 0.5500593185424805,
+      "mean_token_accuracy": 0.8337264291942119,
+      "num_tokens": 2412859.0,
+      "step": 1040
+    },
+    {
+      "epoch": 2.506940253470127,
+      "eval_entropy": 0.6417417074187418,
+      "eval_loss": 0.6651163697242737,
+      "eval_mean_token_accuracy": 0.8159113908081912,
+      "eval_num_tokens": 2412859.0,
+      "eval_runtime": 87.4854,
+      "eval_samples_per_second": 16.231,
+      "eval_steps_per_second": 2.035,
+      "step": 1040
+    },
+    {
+      "entropy": 0.6301592070609331,
+      "epoch": 2.5552202776101387,
+      "grad_norm": 0.8768910765647888,
+      "learning_rate": 0.00034935254927581064,
+      "loss": 0.5613903999328613,
+      "mean_token_accuracy": 0.8298896946012974,
+      "num_tokens": 2458201.0,
+      "step": 1060
+    },
+    {
+      "epoch": 2.5552202776101387,
+      "eval_entropy": 0.6569794751285167,
+      "eval_loss": 0.6668341159820557,
+      "eval_mean_token_accuracy": 0.8126791795987761,
+      "eval_num_tokens": 2458201.0,
+      "eval_runtime": 88.0474,
+      "eval_samples_per_second": 16.128,
+      "eval_steps_per_second": 2.022,
+      "step": 1060
+    },
+    {
+      "entropy": 0.6069021724164486,
+      "epoch": 2.603500301750151,
+      "grad_norm": 0.9033508896827698,
+      "learning_rate": 0.0003476979927158357,
+      "loss": 0.5590654373168945,
+      "mean_token_accuracy": 0.831520090252161,
+      "num_tokens": 2505382.0,
+      "step": 1080
+    },
+    {
+      "epoch": 2.603500301750151,
+      "eval_entropy": 0.6636380777600106,
+      "eval_loss": 0.6596238017082214,
+      "eval_mean_token_accuracy": 0.8177202826135614,
+      "eval_num_tokens": 2505382.0,
+      "eval_runtime": 88.08,
+      "eval_samples_per_second": 16.122,
+      "eval_steps_per_second": 2.021,
+      "step": 1080
+    },
+    {
+      "entropy": 0.6138021353632211,
+      "epoch": 2.651780325890163,
+      "grad_norm": 0.7345808148384094,
+      "learning_rate": 0.0003459982840843664,
+      "loss": 0.5586410522460937,
+      "mean_token_accuracy": 0.8295876093208789,
+      "num_tokens": 2552927.0,
+      "step": 1100
+    },
+    {
+      "epoch": 2.651780325890163,
+      "eval_entropy": 0.5845904127600487,
+      "eval_loss": 0.663780689239502,
+      "eval_mean_token_accuracy": 0.8142095107710763,
+      "eval_num_tokens": 2552927.0,
+      "eval_runtime": 88.0083,
+      "eval_samples_per_second": 16.135,
+      "eval_steps_per_second": 2.023,
+      "step": 1100
+    },
+    {
+      "entropy": 0.6251844003796577,
+      "epoch": 2.700060350030175,
+      "grad_norm": 0.7587347626686096,
+      "learning_rate": 0.00034425390437883976,
+      "loss": 0.5720802307128906,
+      "mean_token_accuracy": 0.8285771444439888,
+      "num_tokens": 2597426.0,
+      "step": 1120
+    },
+    {
+      "epoch": 2.700060350030175,
+      "eval_entropy": 0.6251207221759839,
+      "eval_loss": 0.6556326746940613,
+      "eval_mean_token_accuracy": 0.8181774036937886,
+      "eval_num_tokens": 2597426.0,
+      "eval_runtime": 87.8517,
+      "eval_samples_per_second": 16.164,
+      "eval_steps_per_second": 2.026,
+      "step": 1120
+    }
+  ],
+  "logging_steps": 20,
+  "max_steps": 4150,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 10,
+  "save_steps": 20,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 1.0975532670678835e+17,
+  "train_batch_size": 4,
+  "trial_name": null,
+  "trial_params": null
+}

overgeneralisation_original_Estonian/Qwen3.5-4B-Base_overgeneralisation_splits_original_features_train_overgeneralisation_splits_original_features_test2/checkpoint-1140/README.md

@@ -0,0 +1,209 @@
+---
+base_model: Qwen/Qwen3.5-4B-Base
+library_name: peft
+pipeline_tag: text-generation
+tags:
+- base_model:adapter:Qwen/Qwen3.5-4B-Base
+- lora
+- sft
+- transformers
+- trl
+---
+
+# Model Card for Model ID
+
+<!-- Provide a quick summary of what the model is/does. -->
+
+
+
+## Model Details
+
+### Model Description
+
+<!-- Provide a longer summary of what this model is. -->
+
+
+
+- **Developed by:** [More Information Needed]
+- **Funded by [optional]:** [More Information Needed]
+- **Shared by [optional]:** [More Information Needed]
+- **Model type:** [More Information Needed]
+- **Language(s) (NLP):** [More Information Needed]
+- **License:** [More Information Needed]
+- **Finetuned from model [optional]:** [More Information Needed]
+
+### Model Sources [optional]
+
+<!-- Provide the basic links for the model. -->
+
+- **Repository:** [More Information Needed]
+- **Paper [optional]:** [More Information Needed]
+- **Demo [optional]:** [More Information Needed]
+
+## Uses
+
+<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
+
+### Direct Use
+
+<!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
+
+[More Information Needed]
+
+### Downstream Use [optional]
+
+<!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->
+
+[More Information Needed]
+
+### Out-of-Scope Use
+
+<!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
+
+[More Information Needed]
+
+## Bias, Risks, and Limitations
+
+<!-- This section is meant to convey both technical and sociotechnical limitations. -->
+
+[More Information Needed]
+
+### Recommendations
+
+<!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->
+
+Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
+
+## How to Get Started with the Model
+
+Use the code below to get started with the model.
+
+[More Information Needed]
+
+## Training Details
+
+### Training Data
+
+<!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
+
+[More Information Needed]
+
+### Training Procedure
+
+<!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
+
+#### Preprocessing [optional]
+
+[More Information Needed]
+
+
+#### Training Hyperparameters
+
+- **Training regime:** [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->
+
+#### Speeds, Sizes, Times [optional]
+
+<!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
+
+[More Information Needed]
+
+## Evaluation
+
+<!-- This section describes the evaluation protocols and provides the results. -->
+
+### Testing Data, Factors & Metrics
+
+#### Testing Data
+
+<!-- This should link to a Dataset Card if possible. -->
+
+[More Information Needed]
+
+#### Factors
+
+<!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->
+
+[More Information Needed]
+
+#### Metrics
+
+<!-- These are the evaluation metrics being used, ideally with a description of why. -->
+
+[More Information Needed]
+
+### Results
+
+[More Information Needed]
+
+#### Summary
+
+
+
+## Model Examination [optional]
+
+<!-- Relevant interpretability work for the model goes here -->
+
+[More Information Needed]
+
+## Environmental Impact
+
+<!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly -->
+
+Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700).
+
+- **Hardware Type:** [More Information Needed]
+- **Hours used:** [More Information Needed]
+- **Cloud Provider:** [More Information Needed]
+- **Compute Region:** [More Information Needed]
+- **Carbon Emitted:** [More Information Needed]
+
+## Technical Specifications [optional]
+
+### Model Architecture and Objective
+
+[More Information Needed]
+
+### Compute Infrastructure
+
+[More Information Needed]
+
+#### Hardware
+
+[More Information Needed]
+
+#### Software
+
+[More Information Needed]
+
+## Citation [optional]
+
+<!-- If there is a paper or blog post introducing the model, the APA and Bibtex information for that should go in this section. -->
+
+**BibTeX:**
+
+[More Information Needed]
+
+**APA:**
+
+[More Information Needed]
+
+## Glossary [optional]
+
+<!-- If relevant, include terms and calculations in this section that can help readers understand the model or model card. -->
+
+[More Information Needed]
+
+## More Information [optional]
+
+[More Information Needed]
+
+## Model Card Authors [optional]
+
+[More Information Needed]
+
+## Model Card Contact
+
+[More Information Needed]
+### Framework versions
+
+- PEFT 0.18.1

overgeneralisation_original_Estonian/Qwen3.5-4B-Base_overgeneralisation_splits_original_features_train_overgeneralisation_splits_original_features_test2/checkpoint-1140/adapter_config.json

@@ -0,0 +1,46 @@
+{
+  "alora_invocation_tokens": null,
+  "alpha_pattern": {},
+  "arrow_config": null,
+  "auto_mapping": null,
+  "base_model_name_or_path": "Qwen/Qwen3.5-4B-Base",
+  "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": 128,
+  "lora_bias": false,
+  "lora_dropout": 0.03828026524568501,
+  "megatron_config": null,
+  "megatron_core": "megatron.core",
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "peft_version": "0.18.1",
+  "qalora_group_size": 16,
+  "r": 64,
+  "rank_pattern": {},
+  "revision": null,
+  "target_modules": [
+    "gate_proj",
+    "v_proj",
+    "up_proj",
+    "down_proj",
+    "o_proj",
+    "q_proj",
+    "k_proj"
+  ],
+  "target_parameters": null,
+  "task_type": "CAUSAL_LM",
+  "trainable_token_indices": null,
+  "use_dora": false,
+  "use_qalora": false,
+  "use_rslora": false
+}

overgeneralisation_original_Estonian/Qwen3.5-4B-Base_overgeneralisation_splits_original_features_train_overgeneralisation_splits_original_features_test2/checkpoint-1140/chat_template.jinja

@@ -0,0 +1,154 @@
+{%- set image_count = namespace(value=0) %}
+{%- set video_count = namespace(value=0) %}
+{%- macro render_content(content, do_vision_count, is_system_content=false) %}
+    {%- if content is string %}
+        {{- content }}
+    {%- elif content is iterable and content is not mapping %}
+        {%- for item in content %}
+            {%- if 'image' in item or 'image_url' in item or item.type == 'image' %}
+                {%- if is_system_content %}
+                    {{- raise_exception('System message cannot contain images.') }}
+                {%- endif %}
+                {%- if do_vision_count %}
+                    {%- set image_count.value = image_count.value + 1 %}
+                {%- endif %}
+                {%- if add_vision_id %}
+                    {{- 'Picture ' ~ image_count.value ~ ': ' }}
+                {%- endif %}
+                {{- '<|vision_start|><|image_pad|><|vision_end|>' }}
+            {%- elif 'video' in item or item.type == 'video' %}
+                {%- if is_system_content %}
+                    {{- raise_exception('System message cannot contain videos.') }}
+                {%- endif %}
+                {%- if do_vision_count %}
+                    {%- set video_count.value = video_count.value + 1 %}
+                {%- endif %}
+                {%- if add_vision_id %}
+                    {{- 'Video ' ~ video_count.value ~ ': ' }}
+                {%- endif %}
+                {{- '<|vision_start|><|video_pad|><|vision_end|>' }}
+            {%- elif 'text' in item %}
+                {{- item.text }}
+            {%- else %}
+                {{- raise_exception('Unexpected item type in content.') }}
+            {%- endif %}
+        {%- endfor %}
+    {%- elif content is none or content is undefined %}
+        {{- '' }}
+    {%- else %}
+        {{- raise_exception('Unexpected content type.') }}
+    {%- endif %}
+{%- endmacro %}
+{%- if not messages %}
+    {{- raise_exception('No messages provided.') }}
+{%- endif %}
+{%- if tools and tools is iterable and tools is not mapping %}
+    {{- '<|im_start|>system\n' }}
+    {{- "# Tools\n\nYou have access to the following functions:\n\n<tools>" }}
+    {%- for tool in tools %}
+        {{- "\n" }}
+        {{- tool | tojson }}
+    {%- endfor %}
+    {{- "\n</tools>" }}
+    {{- '\n\nIf you choose to call a function ONLY reply in the following format with NO suffix:\n\n<tool_call>\n<function=example_function_name>\n<parameter=example_parameter_1>\nvalue_1\n</parameter>\n<parameter=example_parameter_2>\nThis is the value for the second parameter\nthat can span\nmultiple lines\n</parameter>\n</function>\n</tool_call>\n\n<IMPORTANT>\nReminder:\n- Function calls MUST follow the specified format: an inner <function=...></function> block must be nested within <tool_call></tool_call> XML tags\n- Required parameters MUST be specified\n- You may provide optional reasoning for your function call in natural language BEFORE the function call, but NOT after\n- If there is no function call available, answer the question like normal with your current knowledge and do not tell the user about function calls\n</IMPORTANT>' }}
+    {%- if messages[0].role == 'system' %}
+        {%- set content = render_content(messages[0].content, false, true)|trim %}
+        {%- if content %}
+            {{- '\n\n' + content }}
+        {%- endif %}
+    {%- endif %}
+    {{- '<|im_end|>\n' }}
+{%- else %}
+    {%- if messages[0].role == 'system' %}
+        {%- set content = render_content(messages[0].content, false, true)|trim %}
+        {{- '<|im_start|>system\n' + content + '<|im_end|>\n' }}
+    {%- endif %}
+{%- endif %}
+{%- set ns = namespace(multi_step_tool=true, last_query_index=messages|length - 1) %}
+{%- for message in messages[::-1] %}
+    {%- set index = (messages|length - 1) - loop.index0 %}
+    {%- if ns.multi_step_tool and message.role == "user" %}
+        {%- set content = render_content(message.content, false)|trim %}
+        {%- if not(content.startswith('<tool_response>') and content.endswith('</tool_response>')) %}
+            {%- set ns.multi_step_tool = false %}
+            {%- set ns.last_query_index = index %}
+        {%- endif %}
+    {%- endif %}
+{%- endfor %}
+{%- if ns.multi_step_tool %}
+    {{- raise_exception('No user query found in messages.') }}
+{%- endif %}
+{%- for message in messages %}
+    {%- set content = render_content(message.content, true)|trim %}
+    {%- if message.role == "system" %}
+        {%- if not loop.first %}
+            {{- raise_exception('System message must be at the beginning.') }}
+        {%- endif %}
+    {%- elif message.role == "user" %}
+        {{- '<|im_start|>' + message.role + '\n' + content + '<|im_end|>' + '\n' }}
+    {%- elif message.role == "assistant" %}
+        {%- set reasoning_content = '' %}
+        {%- if message.reasoning_content is string %}
+            {%- set reasoning_content = message.reasoning_content %}
+        {%- else %}
+            {%- if '</think>' in content %}
+                {%- set reasoning_content = content.split('</think>')[0].rstrip('\n').split('<think>')[-1].lstrip('\n') %}
+                {%- set content = content.split('</think>')[-1].lstrip('\n') %}
+            {%- endif %}
+        {%- endif %}
+        {%- set reasoning_content = reasoning_content|trim %}
+        {%- if loop.index0 > ns.last_query_index %}
+            {{- '<|im_start|>' + message.role + '\n<think>\n' + reasoning_content + '\n</think>\n\n' + content }}
+        {%- else %}
+            {{- '<|im_start|>' + message.role + '\n' + content }}
+        {%- endif %}
+        {%- if message.tool_calls and message.tool_calls is iterable and message.tool_calls is not mapping %}
+            {%- for tool_call in message.tool_calls %}
+                {%- if tool_call.function is defined %}
+                    {%- set tool_call = tool_call.function %}
+                {%- endif %}
+                {%- if loop.first %}
+                    {%- if content|trim %}
+                        {{- '\n\n<tool_call>\n<function=' + tool_call.name + '>\n' }}
+                    {%- else %}
+                        {{- '<tool_call>\n<function=' + tool_call.name + '>\n' }}
+                    {%- endif %}
+                {%- else %}
+                    {{- '\n<tool_call>\n<function=' + tool_call.name + '>\n' }}
+                {%- endif %}
+                {%- if tool_call.arguments is defined %}
+                    {%- for args_name, args_value in tool_call.arguments|items %}
+                        {{- '<parameter=' + args_name + '>\n' }}
+                        {%- set args_value = args_value | tojson | safe if args_value is mapping or (args_value is sequence and args_value is not string) else args_value | string %}
+                        {{- args_value }}
+                        {{- '\n</parameter>\n' }}
+                    {%- endfor %}
+                {%- endif %}
+                {{- '</function>\n</tool_call>' }}
+            {%- endfor %}
+        {%- endif %}
+        {{- '<|im_end|>\n' }}
+    {%- elif message.role == "tool" %}
+        {%- if loop.previtem and loop.previtem.role != "tool" %}
+            {{- '<|im_start|>user' }}
+        {%- endif %}
+        {{- '\n<tool_response>\n' }}
+        {{- content }}
+        {{- '\n</tool_response>' }}
+        {%- if not loop.last and loop.nextitem.role != "tool" %}
+            {{- '<|im_end|>\n' }}
+        {%- elif loop.last %}
+            {{- '<|im_end|>\n' }}
+        {%- endif %}
+    {%- else %}
+        {{- raise_exception('Unexpected message role.') }}
+    {%- endif %}
+{%- endfor %}
+{%- if add_generation_prompt %}
+    {{- '<|im_start|>assistant\n' }}
+    {%- if enable_thinking is defined and enable_thinking is false %}
+        {{- '<think>\n\n</think>\n\n' }}
+    {%- else %}
+        {{- '<think>\n' }}
+    {%- endif %}
+{%- endif %}

overgeneralisation_original_Estonian/Qwen3.5-4B-Base_overgeneralisation_splits_original_features_train_overgeneralisation_splits_original_features_test2/checkpoint-1140/tokenizer_config.json

@@ -0,0 +1,31 @@
+{
+  "add_prefix_space": false,
+  "audio_bos_token": "<|audio_start|>",
+  "audio_eos_token": "<|audio_end|>",
+  "audio_token": "<|audio_pad|>",
+  "backend": "tokenizers",
+  "bos_token": null,
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<|endoftext|>",
+  "errors": "replace",
+  "image_token": "<|image_pad|>",
+  "is_local": false,
+  "model_max_length": 262144,
+  "model_specific_special_tokens": {
+    "audio_bos_token": "<|audio_start|>",
+    "audio_eos_token": "<|audio_end|>",
+    "audio_token": "<|audio_pad|>",
+    "image_token": "<|image_pad|>",
+    "video_token": "<|video_pad|>",
+    "vision_bos_token": "<|vision_start|>",
+    "vision_eos_token": "<|vision_end|>"
+  },
+  "pad_token": "<|endoftext|>",
+  "pretokenize_regex": "(?i:'s|'t|'re|'ve|'m|'ll|'d)|[^\\r\\n\\p{L}\\p{N}]?[\\p{L}\\p{M}]+|\\p{N}| ?[^\\s\\p{L}\\p{M}\\p{N}]+[\\r\\n]*|\\s*[\\r\\n]+|\\s+(?!\\S)|\\s+",
+  "split_special_tokens": false,
+  "tokenizer_class": "TokenizersBackend",
+  "unk_token": null,
+  "video_token": "<|video_pad|>",
+  "vision_bos_token": "<|vision_start|>",
+  "vision_eos_token": "<|vision_end|>"
+}

overgeneralisation_original_Estonian/Qwen3.5-4B-Base_overgeneralisation_splits_original_features_train_overgeneralisation_splits_original_features_test2/checkpoint-1140/trainer_state.json

@@ -0,0 +1,1231 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 2.748340374170187,
+  "eval_steps": 20,
+  "global_step": 1140,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "entropy": 2.172999547421932,
+      "epoch": 0.04828002414001207,
+      "grad_norm": 2.5071208477020264,
+      "learning_rate": 1.7227585487708594e-05,
+      "loss": 1.9347333908081055,
+      "mean_token_accuracy": 0.6086816020309925,
+      "num_tokens": 47778.0,
+      "step": 20
+    },
+    {
+      "epoch": 0.04828002414001207,
+      "eval_entropy": 1.6909754791956269,
+      "eval_loss": 1.4728385210037231,
+      "eval_mean_token_accuracy": 0.6773593797442619,
+      "eval_num_tokens": 47778.0,
+      "eval_runtime": 90.1265,
+      "eval_samples_per_second": 15.756,
+      "eval_steps_per_second": 1.975,
+      "step": 20
+    },
+    {
+      "entropy": 1.2014626950025558,
+      "epoch": 0.09656004828002414,
+      "grad_norm": 1.5074543952941895,
+      "learning_rate": 3.5361886001086065e-05,
+      "loss": 1.0867146492004394,
+      "mean_token_accuracy": 0.7285927519202232,
+      "num_tokens": 95981.0,
+      "step": 40
+    },
+    {
+      "epoch": 0.09656004828002414,
+      "eval_entropy": 1.0304168352250302,
+      "eval_loss": 0.9211422204971313,
+      "eval_mean_token_accuracy": 0.7613122848312507,
+      "eval_num_tokens": 95981.0,
+      "eval_runtime": 88.6338,
+      "eval_samples_per_second": 16.021,
+      "eval_steps_per_second": 2.008,
+      "step": 40
+    },
+    {
+      "entropy": 0.9454198732972146,
+      "epoch": 0.14484007242003621,
+      "grad_norm": 1.7578108310699463,
+      "learning_rate": 5.349618651446354e-05,
+      "loss": 0.8447297096252442,
+      "mean_token_accuracy": 0.7723424732685089,
+      "num_tokens": 142784.0,
+      "step": 60
+    },
+    {
+      "epoch": 0.14484007242003621,
+      "eval_entropy": 0.8647834474451086,
+      "eval_loss": 0.8240737318992615,
+      "eval_mean_token_accuracy": 0.7821227014064789,
+      "eval_num_tokens": 142784.0,
+      "eval_runtime": 88.0135,
+      "eval_samples_per_second": 16.134,
+      "eval_steps_per_second": 2.022,
+      "step": 60
+    },
+    {
+      "entropy": 0.8963550612330436,
+      "epoch": 0.19312009656004828,
+      "grad_norm": 1.1464931964874268,
+      "learning_rate": 7.1630487027841e-05,
+      "loss": 0.7979836463928223,
+      "mean_token_accuracy": 0.7822489373385906,
+      "num_tokens": 187338.0,
+      "step": 80
+    },
+    {
+      "epoch": 0.19312009656004828,
+      "eval_entropy": 0.8663296133614657,
+      "eval_loss": 0.7752988338470459,
+      "eval_mean_token_accuracy": 0.7885957012015782,
+      "eval_num_tokens": 187338.0,
+      "eval_runtime": 88.284,
+      "eval_samples_per_second": 16.084,
+      "eval_steps_per_second": 2.016,
+      "step": 80
+    },
+    {
+      "entropy": 0.8265103206038475,
+      "epoch": 0.24140012070006034,
+      "grad_norm": 1.0106662511825562,
+      "learning_rate": 8.976478754121848e-05,
+      "loss": 0.7297060012817382,
+      "mean_token_accuracy": 0.7955138415098191,
+      "num_tokens": 236038.0,
+      "step": 100
+    },
+    {
+      "epoch": 0.24140012070006034,
+      "eval_entropy": 0.8122169536151244,
+      "eval_loss": 0.754906177520752,
+      "eval_mean_token_accuracy": 0.7917155255092664,
+      "eval_num_tokens": 236038.0,
+      "eval_runtime": 87.5754,
+      "eval_samples_per_second": 16.215,
+      "eval_steps_per_second": 2.033,
+      "step": 100
+    },
+    {
+      "entropy": 0.8098392844200134,
+      "epoch": 0.28968014484007243,
+      "grad_norm": 1.2238041162490845,
+      "learning_rate": 0.00010789908805459595,
+      "loss": 0.7201489448547364,
+      "mean_token_accuracy": 0.7975016921758652,
+      "num_tokens": 282918.0,
+      "step": 120
+    },
+    {
+      "epoch": 0.28968014484007243,
+      "eval_entropy": 0.8418726462326692,
+      "eval_loss": 0.7393178939819336,
+      "eval_mean_token_accuracy": 0.7936771585700217,
+      "eval_num_tokens": 282918.0,
+      "eval_runtime": 87.1597,
+      "eval_samples_per_second": 16.292,
+      "eval_steps_per_second": 2.042,
+      "step": 120
+    },
+    {
+      "entropy": 0.8080591082572937,
+      "epoch": 0.33796016898008446,
+      "grad_norm": 1.011591911315918,
+      "learning_rate": 0.00012603338856797343,
+      "loss": 0.7147697925567627,
+      "mean_token_accuracy": 0.7947175532579422,
+      "num_tokens": 331305.0,
+      "step": 140
+    },
+    {
+      "epoch": 0.33796016898008446,
+      "eval_entropy": 0.7894941431083037,
+      "eval_loss": 0.720219075679779,
+      "eval_mean_token_accuracy": 0.801899236239744,
+      "eval_num_tokens": 331305.0,
+      "eval_runtime": 87.1493,
+      "eval_samples_per_second": 16.294,
+      "eval_steps_per_second": 2.042,
+      "step": 140
+    },
+    {
+      "entropy": 0.7957205109298229,
+      "epoch": 0.38624019312009655,
+      "grad_norm": 0.8481830954551697,
+      "learning_rate": 0.00014416768908135088,
+      "loss": 0.7050027847290039,
+      "mean_token_accuracy": 0.7995368830859662,
+      "num_tokens": 375876.0,
+      "step": 160
+    },
+    {
+      "epoch": 0.38624019312009655,
+      "eval_entropy": 0.7746140591883928,
+      "eval_loss": 0.7175475358963013,
+      "eval_mean_token_accuracy": 0.7994848955213354,
+      "eval_num_tokens": 375876.0,
+      "eval_runtime": 86.655,
+      "eval_samples_per_second": 16.387,
+      "eval_steps_per_second": 2.054,
+      "step": 160
+    },
+    {
+      "entropy": 0.779565304517746,
+      "epoch": 0.43452021726010864,
+      "grad_norm": 1.032624363899231,
+      "learning_rate": 0.00016230198959472835,
+      "loss": 0.6978332042694092,
+      "mean_token_accuracy": 0.8015618294477462,
+      "num_tokens": 423913.0,
+      "step": 180
+    },
+    {
+      "epoch": 0.43452021726010864,
+      "eval_entropy": 0.7455583871080634,
+      "eval_loss": 0.7118850350379944,
+      "eval_mean_token_accuracy": 0.7999682195400923,
+      "eval_num_tokens": 423913.0,
+      "eval_runtime": 86.7842,
+      "eval_samples_per_second": 16.362,
+      "eval_steps_per_second": 2.051,
+      "step": 180
+    },
+    {
+      "entropy": 0.7810172818601131,
+      "epoch": 0.4828002414001207,
+      "grad_norm": 0.9276663064956665,
+      "learning_rate": 0.00018043629010810582,
+      "loss": 0.6937861442565918,
+      "mean_token_accuracy": 0.802893053740263,
+      "num_tokens": 472395.0,
+      "step": 200
+    },
+    {
+      "epoch": 0.4828002414001207,
+      "eval_entropy": 0.7650193290764027,
+      "eval_loss": 0.7020567059516907,
+      "eval_mean_token_accuracy": 0.8045302153973097,
+      "eval_num_tokens": 472395.0,
+      "eval_runtime": 88.326,
+      "eval_samples_per_second": 16.077,
+      "eval_steps_per_second": 2.015,
+      "step": 200
+    },
+    {
+      "entropy": 0.7862060204148292,
+      "epoch": 0.5310802655401328,
+      "grad_norm": 0.8195295929908752,
+      "learning_rate": 0.0001985705906214833,
+      "loss": 0.6878782272338867,
+      "mean_token_accuracy": 0.8027168907225132,
+      "num_tokens": 517049.0,
+      "step": 220
+    },
+    {
+      "epoch": 0.5310802655401328,
+      "eval_entropy": 0.7656565583154057,
+      "eval_loss": 0.6973932981491089,
+      "eval_mean_token_accuracy": 0.8047967278555538,
+      "eval_num_tokens": 517049.0,
+      "eval_runtime": 87.7698,
+      "eval_samples_per_second": 16.179,
+      "eval_steps_per_second": 2.028,
+      "step": 220
+    },
+    {
+      "entropy": 0.7611534461379051,
+      "epoch": 0.5793602896801449,
+      "grad_norm": 0.7499712109565735,
+      "learning_rate": 0.00021670489113486077,
+      "loss": 0.681937837600708,
+      "mean_token_accuracy": 0.8041978515684605,
+      "num_tokens": 562836.0,
+      "step": 240
+    },
+    {
+      "epoch": 0.5793602896801449,
+      "eval_entropy": 0.7599442380197933,
+      "eval_loss": 0.7002046704292297,
+      "eval_mean_token_accuracy": 0.8051861517884759,
+      "eval_num_tokens": 562836.0,
+      "eval_runtime": 87.9239,
+      "eval_samples_per_second": 16.15,
+      "eval_steps_per_second": 2.024,
+      "step": 240
+    },
+    {
+      "entropy": 0.7800783462822437,
+      "epoch": 0.627640313820157,
+      "grad_norm": 0.8645269274711609,
+      "learning_rate": 0.00023483919164823825,
+      "loss": 0.6954500675201416,
+      "mean_token_accuracy": 0.8031484372913837,
+      "num_tokens": 606612.0,
+      "step": 260
+    },
+    {
+      "epoch": 0.627640313820157,
+      "eval_entropy": 0.7498679090751691,
+      "eval_loss": 0.7046149373054504,
+      "eval_mean_token_accuracy": 0.802388215667746,
+      "eval_num_tokens": 606612.0,
+      "eval_runtime": 88.2468,
+      "eval_samples_per_second": 16.091,
+      "eval_steps_per_second": 2.017,
+      "step": 260
+    },
+    {
+      "entropy": 0.7586533665657044,
+      "epoch": 0.6759203379601689,
+      "grad_norm": 0.741596519947052,
+      "learning_rate": 0.0002529734921616157,
+      "loss": 0.690476369857788,
+      "mean_token_accuracy": 0.8046923100948333,
+      "num_tokens": 655920.0,
+      "step": 280
+    },
+    {
+      "epoch": 0.6759203379601689,
+      "eval_entropy": 0.7714274066887544,
+      "eval_loss": 0.6992160677909851,
+      "eval_mean_token_accuracy": 0.8042646482419432,
+      "eval_num_tokens": 655920.0,
+      "eval_runtime": 88.1754,
+      "eval_samples_per_second": 16.104,
+      "eval_steps_per_second": 2.019,
+      "step": 280
+    },
+    {
+      "entropy": 0.7585848286747933,
+      "epoch": 0.724200362100181,
+      "grad_norm": 1.0778124332427979,
+      "learning_rate": 0.00027110779267499317,
+      "loss": 0.6790409564971924,
+      "mean_token_accuracy": 0.8060351841151714,
+      "num_tokens": 702638.0,
+      "step": 300
+    },
+    {
+      "epoch": 0.724200362100181,
+      "eval_entropy": 0.7380402811457601,
+      "eval_loss": 0.7054820656776428,
+      "eval_mean_token_accuracy": 0.8003534738267406,
+      "eval_num_tokens": 702638.0,
+      "eval_runtime": 88.238,
+      "eval_samples_per_second": 16.093,
+      "eval_steps_per_second": 2.017,
+      "step": 300
+    },
+    {
+      "entropy": 0.7696981988847256,
+      "epoch": 0.7724803862401931,
+      "grad_norm": 0.8408658504486084,
+      "learning_rate": 0.00028924209318837064,
+      "loss": 0.7070387363433838,
+      "mean_token_accuracy": 0.8035947173833847,
+      "num_tokens": 749377.0,
+      "step": 320
+    },
+    {
+      "epoch": 0.7724803862401931,
+      "eval_entropy": 0.7897409023193831,
+      "eval_loss": 0.7015668749809265,
+      "eval_mean_token_accuracy": 0.805306687113944,
+      "eval_num_tokens": 749377.0,
+      "eval_runtime": 87.9395,
+      "eval_samples_per_second": 16.147,
+      "eval_steps_per_second": 2.024,
+      "step": 320
+    },
+    {
+      "entropy": 0.762337189912796,
+      "epoch": 0.8207604103802052,
+      "grad_norm": 0.9843310117721558,
+      "learning_rate": 0.0003073763937017481,
+      "loss": 0.7021088123321533,
+      "mean_token_accuracy": 0.8029616877436638,
+      "num_tokens": 798874.0,
+      "step": 340
+    },
+    {
+      "epoch": 0.8207604103802052,
+      "eval_entropy": 0.7210077121016685,
+      "eval_loss": 0.7082746624946594,
+      "eval_mean_token_accuracy": 0.8049283824609906,
+      "eval_num_tokens": 798874.0,
+      "eval_runtime": 88.1706,
+      "eval_samples_per_second": 16.105,
+      "eval_steps_per_second": 2.019,
+      "step": 340
+    },
+    {
+      "entropy": 0.7879025422036647,
+      "epoch": 0.8690404345202173,
+      "grad_norm": 0.9713814854621887,
+      "learning_rate": 0.0003255106942151256,
+      "loss": 0.714624547958374,
+      "mean_token_accuracy": 0.79987031519413,
+      "num_tokens": 840825.0,
+      "step": 360
+    },
+    {
+      "epoch": 0.8690404345202173,
+      "eval_entropy": 0.7866930489459735,
+      "eval_loss": 0.7145519852638245,
+      "eval_mean_token_accuracy": 0.8015657988157165,
+      "eval_num_tokens": 840825.0,
+      "eval_runtime": 88.3746,
+      "eval_samples_per_second": 16.068,
+      "eval_steps_per_second": 2.014,
+      "step": 360
+    },
+    {
+      "entropy": 0.7844726584851742,
+      "epoch": 0.9173204586602294,
+      "grad_norm": 1.2028311491012573,
+      "learning_rate": 0.00034364499472850306,
+      "loss": 0.7040786266326904,
+      "mean_token_accuracy": 0.7981885217130185,
+      "num_tokens": 882885.0,
+      "step": 380
+    },
+    {
+      "epoch": 0.9173204586602294,
+      "eval_entropy": 0.7767668339643585,
+      "eval_loss": 0.7073574066162109,
+      "eval_mean_token_accuracy": 0.8040851833445303,
+      "eval_num_tokens": 882885.0,
+      "eval_runtime": 88.3828,
+      "eval_samples_per_second": 16.066,
+      "eval_steps_per_second": 2.014,
+      "step": 380
+    },
+    {
+      "entropy": 0.7699790760874748,
+      "epoch": 0.9656004828002414,
+      "grad_norm": 0.9397398233413696,
+      "learning_rate": 0.00036177929524188054,
+      "loss": 0.7040313720703125,
+      "mean_token_accuracy": 0.8034727744758129,
+      "num_tokens": 927456.0,
+      "step": 400
+    },
+    {
+      "epoch": 0.9656004828002414,
+      "eval_entropy": 0.7749103545472863,
+      "eval_loss": 0.7093513011932373,
+      "eval_mean_token_accuracy": 0.8019482901926791,
+      "eval_num_tokens": 927456.0,
+      "eval_runtime": 86.2112,
+      "eval_samples_per_second": 16.471,
+      "eval_steps_per_second": 2.065,
+      "step": 400
+    },
+    {
+      "entropy": 0.7672389172888422,
+      "epoch": 1.012070006035003,
+      "grad_norm": 0.99210524559021,
+      "learning_rate": 0.00037628567078152296,
+      "loss": 0.6913118362426758,
+      "mean_token_accuracy": 0.8042210944287189,
+      "num_tokens": 972662.0,
+      "step": 420
+    },
+    {
+      "epoch": 1.012070006035003,
+      "eval_entropy": 0.7294771229283193,
+      "eval_loss": 0.7171520590782166,
+      "eval_mean_token_accuracy": 0.8015128585059991,
+      "eval_num_tokens": 972662.0,
+      "eval_runtime": 88.037,
+      "eval_samples_per_second": 16.13,
+      "eval_steps_per_second": 2.022,
+      "step": 420
+    },
+    {
+      "entropy": 0.7143943183124065,
+      "epoch": 1.060350030175015,
+      "grad_norm": 1.087638258934021,
+      "learning_rate": 0.0003762484015601069,
+      "loss": 0.642704439163208,
+      "mean_token_accuracy": 0.8124966286122799,
+      "num_tokens": 1022525.0,
+      "step": 440
+    },
+    {
+      "epoch": 1.060350030175015,
+      "eval_entropy": 0.743511886744017,
+      "eval_loss": 0.7116662859916687,
+      "eval_mean_token_accuracy": 0.8037210672758939,
+      "eval_num_tokens": 1022525.0,
+      "eval_runtime": 87.9931,
+      "eval_samples_per_second": 16.138,
+      "eval_steps_per_second": 2.023,
+      "step": 440
+    },
+    {
+      "entropy": 0.7237346574664116,
+      "epoch": 1.1086300543150271,
+      "grad_norm": 0.9365553259849548,
+      "learning_rate": 0.0003761579008392123,
+      "loss": 0.652644681930542,
+      "mean_token_accuracy": 0.8107496216893196,
+      "num_tokens": 1069194.0,
+      "step": 460
+    },
+    {
+      "epoch": 1.1086300543150271,
+      "eval_entropy": 0.7372823839776972,
+      "eval_loss": 0.7238303422927856,
+      "eval_mean_token_accuracy": 0.8026230124275336,
+      "eval_num_tokens": 1069194.0,
+      "eval_runtime": 87.6778,
+      "eval_samples_per_second": 16.196,
+      "eval_steps_per_second": 2.03,
+      "step": 460
+    },
+    {
+      "entropy": 0.7202137842774391,
+      "epoch": 1.1569100784550392,
+      "grad_norm": 0.7998443841934204,
+      "learning_rate": 0.0003760141942294726,
+      "loss": 0.6485219478607178,
+      "mean_token_accuracy": 0.8092714451253414,
+      "num_tokens": 1117570.0,
+      "step": 480
+    },
+    {
+      "epoch": 1.1569100784550392,
+      "eval_entropy": 0.71547647942318,
+      "eval_loss": 0.7124277949333191,
+      "eval_mean_token_accuracy": 0.8013174219077892,
+      "eval_num_tokens": 1117570.0,
+      "eval_runtime": 87.4387,
+      "eval_samples_per_second": 16.24,
+      "eval_steps_per_second": 2.036,
+      "step": 480
+    },
+    {
+      "entropy": 0.7270819500088692,
+      "epoch": 1.2051901025950513,
+      "grad_norm": 0.9470787644386292,
+      "learning_rate": 0.00037581732239815854,
+      "loss": 0.66352219581604,
+      "mean_token_accuracy": 0.8102593503892421,
+      "num_tokens": 1161717.0,
+      "step": 500
+    },
+    {
+      "epoch": 1.2051901025950513,
+      "eval_entropy": 0.7791565897759427,
+      "eval_loss": 0.7179726958274841,
+      "eval_mean_token_accuracy": 0.7948898328154275,
+      "eval_num_tokens": 1161717.0,
+      "eval_runtime": 87.9927,
+      "eval_samples_per_second": 16.138,
+      "eval_steps_per_second": 2.023,
+      "step": 500
+    },
+    {
+      "entropy": 0.7320936664938926,
+      "epoch": 1.2534701267350634,
+      "grad_norm": 1.0220379829406738,
+      "learning_rate": 0.0003755673410576695,
+      "loss": 0.6581549644470215,
+      "mean_token_accuracy": 0.8098709337413311,
+      "num_tokens": 1209390.0,
+      "step": 520
+    },
+    {
+      "epoch": 1.2534701267350634,
+      "eval_entropy": 0.7710678225822662,
+      "eval_loss": 0.708797812461853,
+      "eval_mean_token_accuracy": 0.8039572778042783,
+      "eval_num_tokens": 1209390.0,
+      "eval_runtime": 88.1193,
+      "eval_samples_per_second": 16.115,
+      "eval_steps_per_second": 2.02,
+      "step": 520
+    },
+    {
+      "entropy": 0.7281674664467573,
+      "epoch": 1.3017501508750755,
+      "grad_norm": 0.8775188326835632,
+      "learning_rate": 0.000375264320949768,
+      "loss": 0.6631278991699219,
+      "mean_token_accuracy": 0.8101261422038079,
+      "num_tokens": 1256611.0,
+      "step": 540
+    },
+    {
+      "epoch": 1.3017501508750755,
+      "eval_entropy": 0.7199799800187014,
+      "eval_loss": 0.6981531381607056,
+      "eval_mean_token_accuracy": 0.8071265254127845,
+      "eval_num_tokens": 1256611.0,
+      "eval_runtime": 88.165,
+      "eval_samples_per_second": 16.106,
+      "eval_steps_per_second": 2.019,
+      "step": 540
+    },
+    {
+      "entropy": 0.722860773652792,
+      "epoch": 1.3500301750150876,
+      "grad_norm": 0.8949288129806519,
+      "learning_rate": 0.00037490834782556,
+      "loss": 0.6639598846435547,
+      "mean_token_accuracy": 0.8105649061501026,
+      "num_tokens": 1302325.0,
+      "step": 560
+    },
+    {
+      "epoch": 1.3500301750150876,
+      "eval_entropy": 0.7134760032209118,
+      "eval_loss": 0.7029948234558105,
+      "eval_mean_token_accuracy": 0.8062985880991046,
+      "eval_num_tokens": 1302325.0,
+      "eval_runtime": 88.1286,
+      "eval_samples_per_second": 16.113,
+      "eval_steps_per_second": 2.02,
+      "step": 560
+    },
+    {
+      "entropy": 0.7299126699566841,
+      "epoch": 1.3983101991550995,
+      "grad_norm": 1.0323039293289185,
+      "learning_rate": 0.0003744995224212291,
+      "loss": 0.6535763263702392,
+      "mean_token_accuracy": 0.8120224848389626,
+      "num_tokens": 1343844.0,
+      "step": 580
+    },
+    {
+      "epoch": 1.3983101991550995,
+      "eval_entropy": 0.7169701182440426,
+      "eval_loss": 0.7067857980728149,
+      "eval_mean_token_accuracy": 0.8040432936689826,
+      "eval_num_tokens": 1343844.0,
+      "eval_runtime": 88.2252,
+      "eval_samples_per_second": 16.095,
+      "eval_steps_per_second": 2.018,
+      "step": 580
+    },
+    {
+      "entropy": 0.7281742256134749,
+      "epoch": 1.4465902232951118,
+      "grad_norm": 1.3370424509048462,
+      "learning_rate": 0.00037403796042952863,
+      "loss": 0.6637272357940673,
+      "mean_token_accuracy": 0.8118261776864528,
+      "num_tokens": 1389471.0,
+      "step": 600
+    },
+    {
+      "epoch": 1.4465902232951118,
+      "eval_entropy": 0.7378843212395572,
+      "eval_loss": 0.7038390040397644,
+      "eval_mean_token_accuracy": 0.8059229371922739,
+      "eval_num_tokens": 1389471.0,
+      "eval_runtime": 87.9301,
+      "eval_samples_per_second": 16.149,
+      "eval_steps_per_second": 2.024,
+      "step": 600
+    },
+    {
+      "entropy": 0.7388614989817143,
+      "epoch": 1.4948702474351236,
+      "grad_norm": 0.8464171290397644,
+      "learning_rate": 0.00037352379246704257,
+      "loss": 0.6604740142822265,
+      "mean_token_accuracy": 0.8102974124252796,
+      "num_tokens": 1435975.0,
+      "step": 620
+    },
+    {
+      "epoch": 1.4948702474351236,
+      "eval_entropy": 0.7126847046814607,
+      "eval_loss": 0.6950703263282776,
+      "eval_mean_token_accuracy": 0.8085885392815879,
+      "eval_num_tokens": 1435975.0,
+      "eval_runtime": 88.2732,
+      "eval_samples_per_second": 16.086,
+      "eval_steps_per_second": 2.016,
+      "step": 620
+    },
+    {
+      "entropy": 0.7258916139602661,
+      "epoch": 1.5431502715751357,
+      "grad_norm": 1.1780085563659668,
+      "learning_rate": 0.0003729571640372223,
+      "loss": 0.6529891014099121,
+      "mean_token_accuracy": 0.8085981778800487,
+      "num_tokens": 1482035.0,
+      "step": 640
+    },
+    {
+      "epoch": 1.5431502715751357,
+      "eval_entropy": 0.7096509638797032,
+      "eval_loss": 0.690051257610321,
+      "eval_mean_token_accuracy": 0.8100520241796301,
+      "eval_num_tokens": 1482035.0,
+      "eval_runtime": 88.3153,
+      "eval_samples_per_second": 16.079,
+      "eval_steps_per_second": 2.016,
+      "step": 640
+    },
+    {
+      "entropy": 0.7151173129677773,
+      "epoch": 1.5914302957151478,
+      "grad_norm": 1.0357667207717896,
+      "learning_rate": 0.0003723382354892108,
+      "loss": 0.6451474189758301,
+      "mean_token_accuracy": 0.8136902332305909,
+      "num_tokens": 1528119.0,
+      "step": 660
+    },
+    {
+      "epoch": 1.5914302957151478,
+      "eval_entropy": 0.7060252652409371,
+      "eval_loss": 0.6899483799934387,
+      "eval_mean_token_accuracy": 0.810834194502134,
+      "eval_num_tokens": 1528119.0,
+      "eval_runtime": 87.8848,
+      "eval_samples_per_second": 16.158,
+      "eval_steps_per_second": 2.025,
+      "step": 660
+    },
+    {
+      "entropy": 0.7141377851366997,
+      "epoch": 1.63971031985516,
+      "grad_norm": 0.8391667008399963,
+      "learning_rate": 0.000371667181972466,
+      "loss": 0.6481579780578614,
+      "mean_token_accuracy": 0.8159015104174614,
+      "num_tokens": 1573661.0,
+      "step": 680
+    },
+    {
+      "epoch": 1.63971031985516,
+      "eval_entropy": 0.7598993319473909,
+      "eval_loss": 0.681204080581665,
+      "eval_mean_token_accuracy": 0.8100194586126992,
+      "eval_num_tokens": 1573661.0,
+      "eval_runtime": 88.2415,
+      "eval_samples_per_second": 16.092,
+      "eval_steps_per_second": 2.017,
+      "step": 680
+    },
+    {
+      "entropy": 0.7235980287194252,
+      "epoch": 1.687990343995172,
+      "grad_norm": 0.8197196125984192,
+      "learning_rate": 0.00037094419338719537,
+      "loss": 0.6582849025726318,
+      "mean_token_accuracy": 0.8114834539592266,
+      "num_tokens": 1621714.0,
+      "step": 700
+    },
+    {
+      "epoch": 1.687990343995172,
+      "eval_entropy": 0.7166854568411795,
+      "eval_loss": 0.679177463054657,
+      "eval_mean_token_accuracy": 0.8119039877077167,
+      "eval_num_tokens": 1621714.0,
+      "eval_runtime": 88.0957,
+      "eval_samples_per_second": 16.119,
+      "eval_steps_per_second": 2.021,
+      "step": 700
+    },
+    {
+      "entropy": 0.7278190158307553,
+      "epoch": 1.736270368135184,
+      "grad_norm": 0.7820602655410767,
+      "learning_rate": 0.0003701694743306164,
+      "loss": 0.6633153438568116,
+      "mean_token_accuracy": 0.8123249113559723,
+      "num_tokens": 1672118.0,
+      "step": 720
+    },
+    {
+      "epoch": 1.736270368135184,
+      "eval_entropy": 0.6980850314826108,
+      "eval_loss": 0.6775653958320618,
+      "eval_mean_token_accuracy": 0.8116872906684875,
+      "eval_num_tokens": 1672118.0,
+      "eval_runtime": 87.9795,
+      "eval_samples_per_second": 16.14,
+      "eval_steps_per_second": 2.023,
+      "step": 720
+    },
+    {
+      "entropy": 0.7149631556123495,
+      "epoch": 1.7845503922751962,
+      "grad_norm": 1.0139760971069336,
+      "learning_rate": 0.00036934324403905795,
+      "loss": 0.6468976020812989,
+      "mean_token_accuracy": 0.8117561548948288,
+      "num_tokens": 1720465.0,
+      "step": 740
+    },
+    {
+      "epoch": 1.7845503922751962,
+      "eval_entropy": 0.69009849094273,
+      "eval_loss": 0.6731093525886536,
+      "eval_mean_token_accuracy": 0.81281964196248,
+      "eval_num_tokens": 1720465.0,
+      "eval_runtime": 87.7412,
+      "eval_samples_per_second": 16.184,
+      "eval_steps_per_second": 2.029,
+      "step": 740
+    },
+    {
+      "entropy": 0.7226049453020096,
+      "epoch": 1.832830416415208,
+      "grad_norm": 1.5040546655654907,
+      "learning_rate": 0.0003684657363259193,
+      "loss": 0.6385328769683838,
+      "mean_token_accuracy": 0.8132639616727829,
+      "num_tokens": 1764440.0,
+      "step": 760
+    },
+    {
+      "epoch": 1.832830416415208,
+      "eval_entropy": 0.7125169618076153,
+      "eval_loss": 0.6864680647850037,
+      "eval_mean_token_accuracy": 0.8049418089095126,
+      "eval_num_tokens": 1764440.0,
+      "eval_runtime": 87.9429,
+      "eval_samples_per_second": 16.147,
+      "eval_steps_per_second": 2.024,
+      "step": 760
+    },
+    {
+      "entropy": 0.707821810990572,
+      "epoch": 1.8811104405552204,
+      "grad_norm": 0.8604223132133484,
+      "learning_rate": 0.00036753719951550327,
+      "loss": 0.6397049427032471,
+      "mean_token_accuracy": 0.8137877315282822,
+      "num_tokens": 1810828.0,
+      "step": 780
+    },
+    {
+      "epoch": 1.8811104405552204,
+      "eval_entropy": 0.6492102551326323,
+      "eval_loss": 0.6743726134300232,
+      "eval_mean_token_accuracy": 0.8130356572317273,
+      "eval_num_tokens": 1810828.0,
+      "eval_runtime": 87.8684,
+      "eval_samples_per_second": 16.161,
+      "eval_steps_per_second": 2.026,
+      "step": 780
+    },
+    {
+      "entropy": 0.6800355084240437,
+      "epoch": 1.9293904646952322,
+      "grad_norm": 0.7270982265472412,
+      "learning_rate": 0.00036655789637274377,
+      "loss": 0.6267755031585693,
+      "mean_token_accuracy": 0.8180945307016373,
+      "num_tokens": 1857600.0,
+      "step": 800
+    },
+    {
+      "epoch": 1.9293904646952322,
+      "eval_entropy": 0.6727538651294922,
+      "eval_loss": 0.6696028709411621,
+      "eval_mean_token_accuracy": 0.8116443090224534,
+      "eval_num_tokens": 1857600.0,
+      "eval_runtime": 88.2449,
+      "eval_samples_per_second": 16.092,
+      "eval_steps_per_second": 2.017,
+      "step": 800
+    },
+    {
+      "entropy": 0.6746691606938839,
+      "epoch": 1.9776704888352445,
+      "grad_norm": 0.9771540760993958,
+      "learning_rate": 0.0003655281040288461,
+      "loss": 0.6326992988586426,
+      "mean_token_accuracy": 0.8179976396262646,
+      "num_tokens": 1901998.0,
+      "step": 820
+    },
+    {
+      "epoch": 1.9776704888352445,
+      "eval_entropy": 0.6636319716324967,
+      "eval_loss": 0.6714363098144531,
+      "eval_mean_token_accuracy": 0.8126635943235976,
+      "eval_num_tokens": 1901998.0,
+      "eval_runtime": 87.7666,
+      "eval_samples_per_second": 16.179,
+      "eval_steps_per_second": 2.028,
+      "step": 820
+    },
+    {
+      "entropy": 0.6319125778489298,
+      "epoch": 2.024140012070006,
+      "grad_norm": 0.8380444645881653,
+      "learning_rate": 0.0003644481139028622,
+      "loss": 0.5780903816223144,
+      "mean_token_accuracy": 0.8250188401767186,
+      "num_tokens": 1947553.0,
+      "step": 840
+    },
+    {
+      "epoch": 2.024140012070006,
+      "eval_entropy": 0.6627151989535,
+      "eval_loss": 0.6774935722351074,
+      "eval_mean_token_accuracy": 0.8140510819601209,
+      "eval_num_tokens": 1947553.0,
+      "eval_runtime": 87.8287,
+      "eval_samples_per_second": 16.168,
+      "eval_steps_per_second": 2.027,
+      "step": 840
+    },
+    {
+      "entropy": 0.6059147048741579,
+      "epoch": 2.0724200362100182,
+      "grad_norm": 0.9745152592658997,
+      "learning_rate": 0.0003633182316192229,
+      "loss": 0.546702527999878,
+      "mean_token_accuracy": 0.8358186542987823,
+      "num_tokens": 1995234.0,
+      "step": 860
+    },
+    {
+      "epoch": 2.0724200362100182,
+      "eval_entropy": 0.6620710384979677,
+      "eval_loss": 0.6769291162490845,
+      "eval_mean_token_accuracy": 0.8135082835561773,
+      "eval_num_tokens": 1995234.0,
+      "eval_runtime": 88.0008,
+      "eval_samples_per_second": 16.136,
+      "eval_steps_per_second": 2.023,
+      "step": 860
+    },
+    {
+      "entropy": 0.6076117843389511,
+      "epoch": 2.12070006035003,
+      "grad_norm": 0.6914380192756653,
+      "learning_rate": 0.0003621387769212491,
+      "loss": 0.5324535846710206,
+      "mean_token_accuracy": 0.8366568259894848,
+      "num_tokens": 2040639.0,
+      "step": 880
+    },
+    {
+      "epoch": 2.12070006035003,
+      "eval_entropy": 0.6592224213514435,
+      "eval_loss": 0.670383632183075,
+      "eval_mean_token_accuracy": 0.8162150088320957,
+      "eval_num_tokens": 2040639.0,
+      "eval_runtime": 88.324,
+      "eval_samples_per_second": 16.077,
+      "eval_steps_per_second": 2.015,
+      "step": 880
+    },
+    {
+      "entropy": 0.6158072650432587,
+      "epoch": 2.1689800844900424,
+      "grad_norm": 0.9649198055267334,
+      "learning_rate": 0.0003609100835806688,
+      "loss": 0.5564568996429443,
+      "mean_token_accuracy": 0.8329654954373836,
+      "num_tokens": 2090604.0,
+      "step": 900
+    },
+    {
+      "epoch": 2.1689800844900424,
+      "eval_entropy": 0.608588819758276,
+      "eval_loss": 0.6745102405548096,
+      "eval_mean_token_accuracy": 0.8155080542135774,
+      "eval_num_tokens": 2090604.0,
+      "eval_runtime": 87.52,
+      "eval_samples_per_second": 16.225,
+      "eval_steps_per_second": 2.034,
+      "step": 900
+    },
+    {
+      "entropy": 0.5938734326511621,
+      "epoch": 2.2172601086300543,
+      "grad_norm": 0.6915557384490967,
+      "learning_rate": 0.0003596324993031632,
+      "loss": 0.5351875782012939,
+      "mean_token_accuracy": 0.8385631121695042,
+      "num_tokens": 2142246.0,
+      "step": 920
+    },
+    {
+      "epoch": 2.2172601086300543,
+      "eval_entropy": 0.6251631205001574,
+      "eval_loss": 0.6664518117904663,
+      "eval_mean_token_accuracy": 0.81711940417129,
+      "eval_num_tokens": 2142246.0,
+      "eval_runtime": 87.1792,
+      "eval_samples_per_second": 16.288,
+      "eval_steps_per_second": 2.042,
+      "step": 920
+    },
+    {
+      "entropy": 0.623173613846302,
+      "epoch": 2.2655401327700666,
+      "grad_norm": 0.8366610407829285,
+      "learning_rate": 0.00035830638562997063,
+      "loss": 0.5488213539123535,
+      "mean_token_accuracy": 0.8321483485400677,
+      "num_tokens": 2183144.0,
+      "step": 940
+    },
+    {
+      "epoch": 2.2655401327700666,
+      "eval_entropy": 0.6237085649136747,
+      "eval_loss": 0.68075031042099,
+      "eval_mean_token_accuracy": 0.814671738429016,
+      "eval_num_tokens": 2183144.0,
+      "eval_runtime": 88.0652,
+      "eval_samples_per_second": 16.124,
+      "eval_steps_per_second": 2.021,
+      "step": 940
+    },
+    {
+      "entropy": 0.6236911740154027,
+      "epoch": 2.3138201569100785,
+      "grad_norm": 0.7812452912330627,
+      "learning_rate": 0.00035693211783557416,
+      "loss": 0.5696200847625732,
+      "mean_token_accuracy": 0.82991396561265,
+      "num_tokens": 2228098.0,
+      "step": 960
+    },
+    {
+      "epoch": 2.3138201569100785,
+      "eval_entropy": 0.6327033426319615,
+      "eval_loss": 0.6742984652519226,
+      "eval_mean_token_accuracy": 0.8148086091105857,
+      "eval_num_tokens": 2228098.0,
+      "eval_runtime": 88.1711,
+      "eval_samples_per_second": 16.105,
+      "eval_steps_per_second": 2.019,
+      "step": 960
+    },
+    {
+      "entropy": 0.6218426413834095,
+      "epoch": 2.3621001810500903,
+      "grad_norm": 0.8719790577888489,
+      "learning_rate": 0.0003555100848215035,
+      "loss": 0.5582651615142822,
+      "mean_token_accuracy": 0.8323961742222309,
+      "num_tokens": 2273062.0,
+      "step": 980
+    },
+    {
+      "epoch": 2.3621001810500903,
+      "eval_entropy": 0.6338012375858393,
+      "eval_loss": 0.6685822606086731,
+      "eval_mean_token_accuracy": 0.8152180300669724,
+      "eval_num_tokens": 2273062.0,
+      "eval_runtime": 88.101,
+      "eval_samples_per_second": 16.118,
+      "eval_steps_per_second": 2.02,
+      "step": 980
+    },
+    {
+      "entropy": 0.6050681818276644,
+      "epoch": 2.4103802051901027,
+      "grad_norm": 0.9151566028594971,
+      "learning_rate": 0.00035404068900628076,
+      "loss": 0.559452486038208,
+      "mean_token_accuracy": 0.8310242861509323,
+      "num_tokens": 2319305.0,
+      "step": 1000
+    },
+    {
+      "epoch": 2.4103802051901027,
+      "eval_entropy": 0.6634890969549672,
+      "eval_loss": 0.6692465543746948,
+      "eval_mean_token_accuracy": 0.8134353673190213,
+      "eval_num_tokens": 2319305.0,
+      "eval_runtime": 87.7711,
+      "eval_samples_per_second": 16.178,
+      "eval_steps_per_second": 2.028,
+      "step": 1000
+    },
+    {
+      "entropy": 0.6169968567788601,
+      "epoch": 2.4586602293301145,
+      "grad_norm": 0.7081006765365601,
+      "learning_rate": 0.0003525243462115402,
+      "loss": 0.5589694023132324,
+      "mean_token_accuracy": 0.8357502184808254,
+      "num_tokens": 2368592.0,
+      "step": 1020
+    },
+    {
+      "epoch": 2.4586602293301145,
+      "eval_entropy": 0.6126393578695447,
+      "eval_loss": 0.6706712245941162,
+      "eval_mean_token_accuracy": 0.8155577500884452,
+      "eval_num_tokens": 2368592.0,
+      "eval_runtime": 88.1407,
+      "eval_samples_per_second": 16.111,
+      "eval_steps_per_second": 2.019,
+      "step": 1020
+    },
+    {
+      "entropy": 0.6072248566895724,
+      "epoch": 2.506940253470127,
+      "grad_norm": 0.6999370455741882,
+      "learning_rate": 0.0003509614855443561,
+      "loss": 0.5500593185424805,
+      "mean_token_accuracy": 0.8337264291942119,
+      "num_tokens": 2412859.0,
+      "step": 1040
+    },
+    {
+      "epoch": 2.506940253470127,
+      "eval_entropy": 0.6417417074187418,
+      "eval_loss": 0.6651163697242737,
+      "eval_mean_token_accuracy": 0.8159113908081912,
+      "eval_num_tokens": 2412859.0,
+      "eval_runtime": 87.4854,
+      "eval_samples_per_second": 16.231,
+      "eval_steps_per_second": 2.035,
+      "step": 1040
+    },
+    {
+      "entropy": 0.6301592070609331,
+      "epoch": 2.5552202776101387,
+      "grad_norm": 0.8768910765647888,
+      "learning_rate": 0.00034935254927581064,
+      "loss": 0.5613903999328613,
+      "mean_token_accuracy": 0.8298896946012974,
+      "num_tokens": 2458201.0,
+      "step": 1060
+    },
+    {
+      "epoch": 2.5552202776101387,
+      "eval_entropy": 0.6569794751285167,
+      "eval_loss": 0.6668341159820557,
+      "eval_mean_token_accuracy": 0.8126791795987761,
+      "eval_num_tokens": 2458201.0,
+      "eval_runtime": 88.0474,
+      "eval_samples_per_second": 16.128,
+      "eval_steps_per_second": 2.022,
+      "step": 1060
+    },
+    {
+      "entropy": 0.6069021724164486,
+      "epoch": 2.603500301750151,
+      "grad_norm": 0.9033508896827698,
+      "learning_rate": 0.0003476979927158357,
+      "loss": 0.5590654373168945,
+      "mean_token_accuracy": 0.831520090252161,
+      "num_tokens": 2505382.0,
+      "step": 1080
+    },
+    {
+      "epoch": 2.603500301750151,
+      "eval_entropy": 0.6636380777600106,
+      "eval_loss": 0.6596238017082214,
+      "eval_mean_token_accuracy": 0.8177202826135614,
+      "eval_num_tokens": 2505382.0,
+      "eval_runtime": 88.08,
+      "eval_samples_per_second": 16.122,
+      "eval_steps_per_second": 2.021,
+      "step": 1080
+    },
+    {
+      "entropy": 0.6138021353632211,
+      "epoch": 2.651780325890163,
+      "grad_norm": 0.7345808148384094,
+      "learning_rate": 0.0003459982840843664,
+      "loss": 0.5586410522460937,
+      "mean_token_accuracy": 0.8295876093208789,
+      "num_tokens": 2552927.0,
+      "step": 1100
+    },
+    {
+      "epoch": 2.651780325890163,
+      "eval_entropy": 0.5845904127600487,
+      "eval_loss": 0.663780689239502,
+      "eval_mean_token_accuracy": 0.8142095107710763,
+      "eval_num_tokens": 2552927.0,
+      "eval_runtime": 88.0083,
+      "eval_samples_per_second": 16.135,
+      "eval_steps_per_second": 2.023,
+      "step": 1100
+    },
+    {
+      "entropy": 0.6251844003796577,
+      "epoch": 2.700060350030175,
+      "grad_norm": 0.7587347626686096,
+      "learning_rate": 0.00034425390437883976,
+      "loss": 0.5720802307128906,
+      "mean_token_accuracy": 0.8285771444439888,
+      "num_tokens": 2597426.0,
+      "step": 1120
+    },
+    {
+      "epoch": 2.700060350030175,
+      "eval_entropy": 0.6251207221759839,
+      "eval_loss": 0.6556326746940613,
+      "eval_mean_token_accuracy": 0.8181774036937886,
+      "eval_num_tokens": 2597426.0,
+      "eval_runtime": 87.8517,
+      "eval_samples_per_second": 16.164,
+      "eval_steps_per_second": 2.026,
+      "step": 1120
+    },
+    {
+      "entropy": 0.6227757651358843,
+      "epoch": 2.748340374170187,
+      "grad_norm": 0.9254975914955139,
+      "learning_rate": 0.00034246534723807843,
+      "loss": 0.5767871856689453,
+      "mean_token_accuracy": 0.8292960874736309,
+      "num_tokens": 2640514.0,
+      "step": 1140
+    },
+    {
+      "epoch": 2.748340374170187,
+      "eval_entropy": 0.6580858331048087,
+      "eval_loss": 0.6579018831253052,
+      "eval_mean_token_accuracy": 0.8178701273510965,
+      "eval_num_tokens": 2640514.0,
+      "eval_runtime": 87.9523,
+      "eval_samples_per_second": 16.145,
+      "eval_steps_per_second": 2.024,
+      "step": 1140
+    }
+  ],
+  "logging_steps": 20,
+  "max_steps": 4150,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 10,
+  "save_steps": 20,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 1.1159542247710925e+17,
+  "train_batch_size": 4,
+  "trial_name": null,
+  "trial_params": null
+}

overgeneralisation_original_Estonian/Qwen3.5-4B-Base_overgeneralisation_splits_original_features_train_overgeneralisation_splits_original_features_test2/checkpoint-1160/README.md

@@ -0,0 +1,209 @@
+---
+base_model: Qwen/Qwen3.5-4B-Base
+library_name: peft
+pipeline_tag: text-generation
+tags:
+- base_model:adapter:Qwen/Qwen3.5-4B-Base
+- lora
+- sft
+- transformers
+- trl
+---
+
+# Model Card for Model ID
+
+<!-- Provide a quick summary of what the model is/does. -->
+
+
+
+## Model Details
+
+### Model Description
+
+<!-- Provide a longer summary of what this model is. -->
+
+
+
+- **Developed by:** [More Information Needed]
+- **Funded by [optional]:** [More Information Needed]
+- **Shared by [optional]:** [More Information Needed]
+- **Model type:** [More Information Needed]
+- **Language(s) (NLP):** [More Information Needed]
+- **License:** [More Information Needed]
+- **Finetuned from model [optional]:** [More Information Needed]
+
+### Model Sources [optional]
+
+<!-- Provide the basic links for the model. -->
+
+- **Repository:** [More Information Needed]
+- **Paper [optional]:** [More Information Needed]
+- **Demo [optional]:** [More Information Needed]
+
+## Uses
+
+<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
+
+### Direct Use
+
+<!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
+
+[More Information Needed]
+
+### Downstream Use [optional]
+
+<!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->
+
+[More Information Needed]
+
+### Out-of-Scope Use
+
+<!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
+
+[More Information Needed]
+
+## Bias, Risks, and Limitations
+
+<!-- This section is meant to convey both technical and sociotechnical limitations. -->
+
+[More Information Needed]
+
+### Recommendations
+
+<!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->
+
+Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
+
+## How to Get Started with the Model
+
+Use the code below to get started with the model.
+
+[More Information Needed]
+
+## Training Details
+
+### Training Data
+
+<!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
+
+[More Information Needed]
+
+### Training Procedure
+
+<!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
+
+#### Preprocessing [optional]
+
+[More Information Needed]
+
+
+#### Training Hyperparameters
+
+- **Training regime:** [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->
+
+#### Speeds, Sizes, Times [optional]
+
+<!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
+
+[More Information Needed]
+
+## Evaluation
+
+<!-- This section describes the evaluation protocols and provides the results. -->
+
+### Testing Data, Factors & Metrics
+
+#### Testing Data
+
+<!-- This should link to a Dataset Card if possible. -->
+
+[More Information Needed]
+
+#### Factors
+
+<!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->
+
+[More Information Needed]
+
+#### Metrics
+
+<!-- These are the evaluation metrics being used, ideally with a description of why. -->
+
+[More Information Needed]
+
+### Results
+
+[More Information Needed]
+
+#### Summary
+
+
+
+## Model Examination [optional]
+
+<!-- Relevant interpretability work for the model goes here -->
+
+[More Information Needed]
+
+## Environmental Impact
+
+<!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly -->
+
+Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700).
+
+- **Hardware Type:** [More Information Needed]
+- **Hours used:** [More Information Needed]
+- **Cloud Provider:** [More Information Needed]
+- **Compute Region:** [More Information Needed]
+- **Carbon Emitted:** [More Information Needed]
+
+## Technical Specifications [optional]
+
+### Model Architecture and Objective
+
+[More Information Needed]
+
+### Compute Infrastructure
+
+[More Information Needed]
+
+#### Hardware
+
+[More Information Needed]
+
+#### Software
+
+[More Information Needed]
+
+## Citation [optional]
+
+<!-- If there is a paper or blog post introducing the model, the APA and Bibtex information for that should go in this section. -->
+
+**BibTeX:**
+
+[More Information Needed]
+
+**APA:**
+
+[More Information Needed]
+
+## Glossary [optional]
+
+<!-- If relevant, include terms and calculations in this section that can help readers understand the model or model card. -->
+
+[More Information Needed]
+
+## More Information [optional]
+
+[More Information Needed]
+
+## Model Card Authors [optional]
+
+[More Information Needed]
+
+## Model Card Contact
+
+[More Information Needed]
+### Framework versions
+
+- PEFT 0.18.1

overgeneralisation_original_Estonian/Qwen3.5-4B-Base_overgeneralisation_splits_original_features_train_overgeneralisation_splits_original_features_test2/checkpoint-1160/adapter_config.json

@@ -0,0 +1,46 @@
+{
+  "alora_invocation_tokens": null,
+  "alpha_pattern": {},
+  "arrow_config": null,
+  "auto_mapping": null,
+  "base_model_name_or_path": "Qwen/Qwen3.5-4B-Base",
+  "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": 128,
+  "lora_bias": false,
+  "lora_dropout": 0.03828026524568501,
+  "megatron_config": null,
+  "megatron_core": "megatron.core",
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "peft_version": "0.18.1",
+  "qalora_group_size": 16,
+  "r": 64,
+  "rank_pattern": {},
+  "revision": null,
+  "target_modules": [
+    "gate_proj",
+    "v_proj",
+    "up_proj",
+    "down_proj",
+    "o_proj",
+    "q_proj",
+    "k_proj"
+  ],
+  "target_parameters": null,
+  "task_type": "CAUSAL_LM",
+  "trainable_token_indices": null,
+  "use_dora": false,
+  "use_qalora": false,
+  "use_rslora": false
+}

overgeneralisation_original_Estonian/Qwen3.5-4B-Base_overgeneralisation_splits_original_features_train_overgeneralisation_splits_original_features_test2/checkpoint-1160/chat_template.jinja

@@ -0,0 +1,154 @@
+{%- set image_count = namespace(value=0) %}
+{%- set video_count = namespace(value=0) %}
+{%- macro render_content(content, do_vision_count, is_system_content=false) %}
+    {%- if content is string %}
+        {{- content }}
+    {%- elif content is iterable and content is not mapping %}
+        {%- for item in content %}
+            {%- if 'image' in item or 'image_url' in item or item.type == 'image' %}
+                {%- if is_system_content %}
+                    {{- raise_exception('System message cannot contain images.') }}
+                {%- endif %}
+                {%- if do_vision_count %}
+                    {%- set image_count.value = image_count.value + 1 %}
+                {%- endif %}
+                {%- if add_vision_id %}
+                    {{- 'Picture ' ~ image_count.value ~ ': ' }}
+                {%- endif %}
+                {{- '<|vision_start|><|image_pad|><|vision_end|>' }}
+            {%- elif 'video' in item or item.type == 'video' %}
+                {%- if is_system_content %}
+                    {{- raise_exception('System message cannot contain videos.') }}
+                {%- endif %}
+                {%- if do_vision_count %}
+                    {%- set video_count.value = video_count.value + 1 %}
+                {%- endif %}
+                {%- if add_vision_id %}
+                    {{- 'Video ' ~ video_count.value ~ ': ' }}
+                {%- endif %}
+                {{- '<|vision_start|><|video_pad|><|vision_end|>' }}
+            {%- elif 'text' in item %}
+                {{- item.text }}
+            {%- else %}
+                {{- raise_exception('Unexpected item type in content.') }}
+            {%- endif %}
+        {%- endfor %}
+    {%- elif content is none or content is undefined %}
+        {{- '' }}
+    {%- else %}
+        {{- raise_exception('Unexpected content type.') }}
+    {%- endif %}
+{%- endmacro %}
+{%- if not messages %}
+    {{- raise_exception('No messages provided.') }}
+{%- endif %}
+{%- if tools and tools is iterable and tools is not mapping %}
+    {{- '<|im_start|>system\n' }}
+    {{- "# Tools\n\nYou have access to the following functions:\n\n<tools>" }}
+    {%- for tool in tools %}
+        {{- "\n" }}
+        {{- tool | tojson }}
+    {%- endfor %}
+    {{- "\n</tools>" }}
+    {{- '\n\nIf you choose to call a function ONLY reply in the following format with NO suffix:\n\n<tool_call>\n<function=example_function_name>\n<parameter=example_parameter_1>\nvalue_1\n</parameter>\n<parameter=example_parameter_2>\nThis is the value for the second parameter\nthat can span\nmultiple lines\n</parameter>\n</function>\n</tool_call>\n\n<IMPORTANT>\nReminder:\n- Function calls MUST follow the specified format: an inner <function=...></function> block must be nested within <tool_call></tool_call> XML tags\n- Required parameters MUST be specified\n- You may provide optional reasoning for your function call in natural language BEFORE the function call, but NOT after\n- If there is no function call available, answer the question like normal with your current knowledge and do not tell the user about function calls\n</IMPORTANT>' }}
+    {%- if messages[0].role == 'system' %}
+        {%- set content = render_content(messages[0].content, false, true)|trim %}
+        {%- if content %}
+            {{- '\n\n' + content }}
+        {%- endif %}
+    {%- endif %}
+    {{- '<|im_end|>\n' }}
+{%- else %}
+    {%- if messages[0].role == 'system' %}
+        {%- set content = render_content(messages[0].content, false, true)|trim %}
+        {{- '<|im_start|>system\n' + content + '<|im_end|>\n' }}
+    {%- endif %}
+{%- endif %}
+{%- set ns = namespace(multi_step_tool=true, last_query_index=messages|length - 1) %}
+{%- for message in messages[::-1] %}
+    {%- set index = (messages|length - 1) - loop.index0 %}
+    {%- if ns.multi_step_tool and message.role == "user" %}
+        {%- set content = render_content(message.content, false)|trim %}
+        {%- if not(content.startswith('<tool_response>') and content.endswith('</tool_response>')) %}
+            {%- set ns.multi_step_tool = false %}
+            {%- set ns.last_query_index = index %}
+        {%- endif %}
+    {%- endif %}
+{%- endfor %}
+{%- if ns.multi_step_tool %}
+    {{- raise_exception('No user query found in messages.') }}
+{%- endif %}
+{%- for message in messages %}
+    {%- set content = render_content(message.content, true)|trim %}
+    {%- if message.role == "system" %}
+        {%- if not loop.first %}
+            {{- raise_exception('System message must be at the beginning.') }}
+        {%- endif %}
+    {%- elif message.role == "user" %}
+        {{- '<|im_start|>' + message.role + '\n' + content + '<|im_end|>' + '\n' }}
+    {%- elif message.role == "assistant" %}
+        {%- set reasoning_content = '' %}
+        {%- if message.reasoning_content is string %}
+            {%- set reasoning_content = message.reasoning_content %}
+        {%- else %}
+            {%- if '</think>' in content %}
+                {%- set reasoning_content = content.split('</think>')[0].rstrip('\n').split('<think>')[-1].lstrip('\n') %}
+                {%- set content = content.split('</think>')[-1].lstrip('\n') %}
+            {%- endif %}
+        {%- endif %}
+        {%- set reasoning_content = reasoning_content|trim %}
+        {%- if loop.index0 > ns.last_query_index %}
+            {{- '<|im_start|>' + message.role + '\n<think>\n' + reasoning_content + '\n</think>\n\n' + content }}
+        {%- else %}
+            {{- '<|im_start|>' + message.role + '\n' + content }}
+        {%- endif %}
+        {%- if message.tool_calls and message.tool_calls is iterable and message.tool_calls is not mapping %}
+            {%- for tool_call in message.tool_calls %}
+                {%- if tool_call.function is defined %}
+                    {%- set tool_call = tool_call.function %}
+                {%- endif %}
+                {%- if loop.first %}
+                    {%- if content|trim %}
+                        {{- '\n\n<tool_call>\n<function=' + tool_call.name + '>\n' }}
+                    {%- else %}
+                        {{- '<tool_call>\n<function=' + tool_call.name + '>\n' }}
+                    {%- endif %}
+                {%- else %}
+                    {{- '\n<tool_call>\n<function=' + tool_call.name + '>\n' }}
+                {%- endif %}
+                {%- if tool_call.arguments is defined %}
+                    {%- for args_name, args_value in tool_call.arguments|items %}
+                        {{- '<parameter=' + args_name + '>\n' }}
+                        {%- set args_value = args_value | tojson | safe if args_value is mapping or (args_value is sequence and args_value is not string) else args_value | string %}
+                        {{- args_value }}
+                        {{- '\n</parameter>\n' }}
+                    {%- endfor %}
+                {%- endif %}
+                {{- '</function>\n</tool_call>' }}
+            {%- endfor %}
+        {%- endif %}
+        {{- '<|im_end|>\n' }}
+    {%- elif message.role == "tool" %}
+        {%- if loop.previtem and loop.previtem.role != "tool" %}
+            {{- '<|im_start|>user' }}
+        {%- endif %}
+        {{- '\n<tool_response>\n' }}
+        {{- content }}
+        {{- '\n</tool_response>' }}
+        {%- if not loop.last and loop.nextitem.role != "tool" %}
+            {{- '<|im_end|>\n' }}
+        {%- elif loop.last %}
+            {{- '<|im_end|>\n' }}
+        {%- endif %}
+    {%- else %}
+        {{- raise_exception('Unexpected message role.') }}
+    {%- endif %}
+{%- endfor %}
+{%- if add_generation_prompt %}
+    {{- '<|im_start|>assistant\n' }}
+    {%- if enable_thinking is defined and enable_thinking is false %}
+        {{- '<think>\n\n</think>\n\n' }}
+    {%- else %}
+        {{- '<think>\n' }}
+    {%- endif %}
+{%- endif %}