Add files using upload-large-folder tool

systematicity_original_Swedish/Qwen3.5-4B-Base_systematicity_splits_original_features_train_systematicity_splits_original_features_test1/checkpoint-1752/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": 0.026477173062682283,
+  "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": [
+    "q_proj",
+    "k_proj",
+    "down_proj",
+    "v_proj",
+    "gate_proj",
+    "o_proj",
+    "up_proj"
+  ],
+  "target_parameters": null,
+  "task_type": "CAUSAL_LM",
+  "trainable_token_indices": null,
+  "use_dora": false,
+  "use_qalora": false,
+  "use_rslora": false
+}

systematicity_original_Swedish/Qwen3.5-4B-Base_systematicity_splits_original_features_train_systematicity_splits_original_features_test1/checkpoint-1752/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 %}

systematicity_original_Swedish/Qwen3.5-4B-Base_systematicity_splits_original_features_train_systematicity_splits_original_features_test1/checkpoint-1752/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|>"
+}

systematicity_original_Swedish/Qwen3.5-4B-Base_systematicity_splits_original_features_train_systematicity_splits_original_features_test1/checkpoint-1752/trainer_state.json

@@ -0,0 +1,428 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 4.0,
+  "eval_steps": 500,
+  "global_step": 1752,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "entropy": 1.4168990224599838,
+      "epoch": 0.11435105774728416,
+      "grad_norm": 1.3415299654006958,
+      "learning_rate": 4.424985043427545e-05,
+      "loss": 1.2160142517089845,
+      "mean_token_accuracy": 0.724469550549984,
+      "num_tokens": 96586.0,
+      "step": 50
+    },
+    {
+      "entropy": 0.7365697559714317,
+      "epoch": 0.22870211549456831,
+      "grad_norm": 0.9586130380630493,
+      "learning_rate": 8.940275904067897e-05,
+      "loss": 0.6445799255371094,
+      "mean_token_accuracy": 0.822787437736988,
+      "num_tokens": 190720.0,
+      "step": 100
+    },
+    {
+      "entropy": 0.691228820681572,
+      "epoch": 0.34305317324185247,
+      "grad_norm": 0.7875016331672668,
+      "learning_rate": 0.0001345556676470825,
+      "loss": 0.6011501312255859,
+      "mean_token_accuracy": 0.8325563898682594,
+      "num_tokens": 288515.0,
+      "step": 150
+    },
+    {
+      "entropy": 0.6706234656274319,
+      "epoch": 0.45740423098913663,
+      "grad_norm": 0.7867940068244934,
+      "learning_rate": 0.000179708576253486,
+      "loss": 0.5852980422973633,
+      "mean_token_accuracy": 0.83610415995121,
+      "num_tokens": 387874.0,
+      "step": 200
+    },
+    {
+      "entropy": 0.6558432649075985,
+      "epoch": 0.5717552887364208,
+      "grad_norm": 0.8049650192260742,
+      "learning_rate": 0.00022486148485988957,
+      "loss": 0.5714554977416992,
+      "mean_token_accuracy": 0.8380795314908027,
+      "num_tokens": 487865.0,
+      "step": 250
+    },
+    {
+      "entropy": 0.6612135827541351,
+      "epoch": 0.6861063464837049,
+      "grad_norm": 0.6776595711708069,
+      "learning_rate": 0.0002700143934662931,
+      "loss": 0.5815717315673828,
+      "mean_token_accuracy": 0.8377441626787185,
+      "num_tokens": 584269.0,
+      "step": 300
+    },
+    {
+      "entropy": 0.6668009972572326,
+      "epoch": 0.8004574042309891,
+      "grad_norm": 1.132367491722107,
+      "learning_rate": 0.0003151673020726966,
+      "loss": 0.5866946411132813,
+      "mean_token_accuracy": 0.8364784654974937,
+      "num_tokens": 676067.0,
+      "step": 350
+    },
+    {
+      "entropy": 0.6667388904094697,
+      "epoch": 0.9148084619782733,
+      "grad_norm": 0.9101859331130981,
+      "learning_rate": 0.0003603202106791001,
+      "loss": 0.5946685028076172,
+      "mean_token_accuracy": 0.8331978982686996,
+      "num_tokens": 774009.0,
+      "step": 400
+    },
+    {
+      "epoch": 1.0,
+      "eval_entropy": 0.5918735759536715,
+      "eval_loss": 0.6536311507225037,
+      "eval_mean_token_accuracy": 0.8276140002438622,
+      "eval_num_tokens": 847606.0,
+      "eval_runtime": 82.2575,
+      "eval_samples_per_second": 13.251,
+      "eval_steps_per_second": 1.666,
+      "step": 438
+    },
+    {
+      "entropy": 0.654371697890577,
+      "epoch": 1.0274442538593482,
+      "grad_norm": 0.923791229724884,
+      "learning_rate": 0.00039553187999846057,
+      "loss": 0.5879265594482422,
+      "mean_token_accuracy": 0.8356293612325252,
+      "num_tokens": 870214.0,
+      "step": 450
+    },
+    {
+      "entropy": 0.6167843751609325,
+      "epoch": 1.1417953116066324,
+      "grad_norm": 0.7943944334983826,
+      "learning_rate": 0.00039530582686207966,
+      "loss": 0.552466049194336,
+      "mean_token_accuracy": 0.8413123995065689,
+      "num_tokens": 971584.0,
+      "step": 500
+    },
+    {
+      "entropy": 0.6443017952144146,
+      "epoch": 1.2561463693539165,
+      "grad_norm": 1.018168568611145,
+      "learning_rate": 0.00039476615968757247,
+      "loss": 0.5773007583618164,
+      "mean_token_accuracy": 0.8377226310968399,
+      "num_tokens": 1063906.0,
+      "step": 550
+    },
+    {
+      "entropy": 0.6407185651361942,
+      "epoch": 1.3704974271012007,
+      "grad_norm": 0.9198986291885376,
+      "learning_rate": 0.00039391373526624385,
+      "loss": 0.572706184387207,
+      "mean_token_accuracy": 0.8367982304096222,
+      "num_tokens": 1159282.0,
+      "step": 600
+    },
+    {
+      "entropy": 0.6494526116549969,
+      "epoch": 1.4848484848484849,
+      "grad_norm": 1.0277528762817383,
+      "learning_rate": 0.00039274990693190134,
+      "loss": 0.578397102355957,
+      "mean_token_accuracy": 0.836915987432003,
+      "num_tokens": 1254998.0,
+      "step": 650
+    },
+    {
+      "entropy": 0.6396139825880528,
+      "epoch": 1.599199542595769,
+      "grad_norm": 0.8996608257293701,
+      "learning_rate": 0.0003912765224122631,
+      "loss": 0.5766008758544922,
+      "mean_token_accuracy": 0.8396538972854615,
+      "num_tokens": 1351877.0,
+      "step": 700
+    },
+    {
+      "entropy": 0.6245273813605309,
+      "epoch": 1.7135506003430532,
+      "grad_norm": 1.1999542713165283,
+      "learning_rate": 0.00038949592089545203,
+      "loss": 0.5718669509887695,
+      "mean_token_accuracy": 0.8406938418745995,
+      "num_tokens": 1448801.0,
+      "step": 750
+    },
+    {
+      "entropy": 0.6350119508802891,
+      "epoch": 1.8279016580903373,
+      "grad_norm": 0.9171736836433411,
+      "learning_rate": 0.00038741092931623213,
+      "loss": 0.5792041778564453,
+      "mean_token_accuracy": 0.8370065066218376,
+      "num_tokens": 1547040.0,
+      "step": 800
+    },
+    {
+      "entropy": 0.6106068694591522,
+      "epoch": 1.9422527158376215,
+      "grad_norm": 0.6869989037513733,
+      "learning_rate": 0.0003850248578678851,
+      "loss": 0.5748995590209961,
+      "mean_token_accuracy": 0.8383069199323654,
+      "num_tokens": 1646444.0,
+      "step": 850
+    },
+    {
+      "epoch": 2.0,
+      "eval_entropy": 0.5696834810893902,
+      "eval_loss": 0.6577129364013672,
+      "eval_mean_token_accuracy": 0.8307611911836332,
+      "eval_num_tokens": 1695212.0,
+      "eval_runtime": 81.951,
+      "eval_samples_per_second": 13.301,
+      "eval_steps_per_second": 1.672,
+      "step": 876
+    },
+    {
+      "entropy": 0.572038494088323,
+      "epoch": 2.0548885077186965,
+      "grad_norm": 0.7493197321891785,
+      "learning_rate": 0.00038234149474685104,
+      "loss": 0.5193290710449219,
+      "mean_token_accuracy": 0.8497188209277119,
+      "num_tokens": 1738782.0,
+      "step": 900
+    },
+    {
+      "entropy": 0.5478819286823273,
+      "epoch": 2.1692395654659804,
+      "grad_norm": 1.2549943923950195,
+      "learning_rate": 0.0003793651001384773,
+      "loss": 0.4793633270263672,
+      "mean_token_accuracy": 0.8575673401355743,
+      "num_tokens": 1837910.0,
+      "step": 950
+    },
+    {
+      "entropy": 0.5431128491461277,
+      "epoch": 2.283590623213265,
+      "grad_norm": 0.8713223338127136,
+      "learning_rate": 0.00037610039945342474,
+      "loss": 0.4849379348754883,
+      "mean_token_accuracy": 0.8562448808550834,
+      "num_tokens": 1937110.0,
+      "step": 1000
+    },
+    {
+      "entropy": 0.5498340383172036,
+      "epoch": 2.3979416809605487,
+      "grad_norm": 0.6592629551887512,
+      "learning_rate": 0.00037255257582546804,
+      "loss": 0.493680419921875,
+      "mean_token_accuracy": 0.854153895676136,
+      "num_tokens": 2036890.0,
+      "step": 1050
+    },
+    {
+      "entropy": 0.5554873237013817,
+      "epoch": 2.512292738707833,
+      "grad_norm": 0.7878748774528503,
+      "learning_rate": 0.0003687272618826019,
+      "loss": 0.4951488494873047,
+      "mean_token_accuracy": 0.8540353265404701,
+      "num_tokens": 2137239.0,
+      "step": 1100
+    },
+    {
+      "entropy": 0.552360344529152,
+      "epoch": 2.6266437964551175,
+      "grad_norm": 0.7380415797233582,
+      "learning_rate": 0.00036463053080451683,
+      "loss": 0.49989810943603513,
+      "mean_token_accuracy": 0.8533406674861908,
+      "num_tokens": 2231081.0,
+      "step": 1150
+    },
+    {
+      "entropy": 0.5345852719247342,
+      "epoch": 2.7409948542024014,
+      "grad_norm": 0.7600613832473755,
+      "learning_rate": 0.0003602688866806424,
+      "loss": 0.5011516189575196,
+      "mean_token_accuracy": 0.8527707865834236,
+      "num_tokens": 2325503.0,
+      "step": 1200
+    },
+    {
+      "entropy": 0.532319468408823,
+      "epoch": 2.8553459119496853,
+      "grad_norm": 0.8197540640830994,
+      "learning_rate": 0.000355649254184066,
+      "loss": 0.5023023605346679,
+      "mean_token_accuracy": 0.8530397671461105,
+      "num_tokens": 2420627.0,
+      "step": 1250
+    },
+    {
+      "entropy": 0.5198138700425625,
+      "epoch": 2.9696969696969697,
+      "grad_norm": 0.8660516142845154,
+      "learning_rate": 0.00035077896757771987,
+      "loss": 0.4941487121582031,
+      "mean_token_accuracy": 0.855004439651966,
+      "num_tokens": 2518051.0,
+      "step": 1300
+    },
+    {
+      "epoch": 3.0,
+      "eval_entropy": 0.5642598084289662,
+      "eval_loss": 0.6125267744064331,
+      "eval_mean_token_accuracy": 0.8375998204641969,
+      "eval_num_tokens": 2542818.0,
+      "eval_runtime": 81.5524,
+      "eval_samples_per_second": 13.366,
+      "eval_steps_per_second": 1.68,
+      "step": 1314
+    },
+    {
+      "entropy": 0.45835615868495805,
+      "epoch": 3.0823327615780447,
+      "grad_norm": 0.734519362449646,
+      "learning_rate": 0.00034566575907029245,
+      "loss": 0.4334117889404297,
+      "mean_token_accuracy": 0.8698135275526095,
+      "num_tokens": 2613861.0,
+      "step": 1350
+    },
+    {
+      "entropy": 0.44648877054452896,
+      "epoch": 3.1966838193253286,
+      "grad_norm": 0.687981128692627,
+      "learning_rate": 0.00034031774654034867,
+      "loss": 0.40152584075927733,
+      "mean_token_accuracy": 0.8759277358651161,
+      "num_tokens": 2709451.0,
+      "step": 1400
+    },
+    {
+      "entropy": 0.4593352344632149,
+      "epoch": 3.311034877072613,
+      "grad_norm": 0.6310222148895264,
+      "learning_rate": 0.0003347434206481497,
+      "loss": 0.41885005950927734,
+      "mean_token_accuracy": 0.8714999267458916,
+      "num_tokens": 2801336.0,
+      "step": 1450
+    },
+    {
+      "entropy": 0.46372457168996334,
+      "epoch": 3.425385934819897,
+      "grad_norm": 0.6795068979263306,
+      "learning_rate": 0.0003289516313556339,
+      "loss": 0.4194014358520508,
+      "mean_token_accuracy": 0.8709167090058326,
+      "num_tokens": 2898945.0,
+      "step": 1500
+    },
+    {
+      "entropy": 0.4548304344713688,
+      "epoch": 3.5397369925671813,
+      "grad_norm": 0.9671820402145386,
+      "learning_rate": 0.00032295157387595997,
+      "loss": 0.4178035736083984,
+      "mean_token_accuracy": 0.8710319435596466,
+      "num_tokens": 2993378.0,
+      "step": 1550
+    },
+    {
+      "entropy": 0.4532127542793751,
+      "epoch": 3.6540880503144653,
+      "grad_norm": 0.7656959295272827,
+      "learning_rate": 0.0003167527740749187,
+      "loss": 0.42567527770996094,
+      "mean_token_accuracy": 0.869033499956131,
+      "num_tokens": 3091323.0,
+      "step": 1600
+    },
+    {
+      "entropy": 0.4460975781828165,
+      "epoch": 3.7684391080617496,
+      "grad_norm": 0.7811372876167297,
+      "learning_rate": 0.0003103650733473916,
+      "loss": 0.4148756790161133,
+      "mean_token_accuracy": 0.8709280017018318,
+      "num_tokens": 3193194.0,
+      "step": 1650
+    },
+    {
+      "entropy": 0.47125090911984446,
+      "epoch": 3.8827901658090336,
+      "grad_norm": 0.8664485216140747,
+      "learning_rate": 0.00030379861299286587,
+      "loss": 0.4209564590454102,
+      "mean_token_accuracy": 0.8695067319273949,
+      "num_tokens": 3292840.0,
+      "step": 1700
+    },
+    {
+      "entropy": 0.4641740974783897,
+      "epoch": 3.997141223556318,
+      "grad_norm": 0.85616534948349,
+      "learning_rate": 0.00029706381811481263,
+      "loss": 0.4319478988647461,
+      "mean_token_accuracy": 0.8689105224609375,
+      "num_tokens": 3387511.0,
+      "step": 1750
+    },
+    {
+      "epoch": 4.0,
+      "eval_entropy": 0.4623802898138979,
+      "eval_loss": 0.6707634925842285,
+      "eval_mean_token_accuracy": 0.8327476525828786,
+      "eval_num_tokens": 3390424.0,
+      "eval_runtime": 81.9106,
+      "eval_samples_per_second": 13.307,
+      "eval_steps_per_second": 1.673,
+      "step": 1752
+    }
+  ],
+  "logging_steps": 50,
+  "max_steps": 4380,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 10,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 1.4076544045638451e+17,
+  "train_batch_size": 4,
+  "trial_name": null,
+  "trial_params": null
+}

systematicity_original_Swedish/Qwen3.5-4B-Base_systematicity_splits_original_features_train_systematicity_splits_original_features_test1/checkpoint-2190/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

systematicity_original_Swedish/Qwen3.5-4B-Base_systematicity_splits_original_features_train_systematicity_splits_original_features_test1/checkpoint-2190/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": 0.026477173062682283,
+  "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": [
+    "q_proj",
+    "k_proj",
+    "down_proj",
+    "v_proj",
+    "gate_proj",
+    "o_proj",
+    "up_proj"
+  ],
+  "target_parameters": null,
+  "task_type": "CAUSAL_LM",
+  "trainable_token_indices": null,
+  "use_dora": false,
+  "use_qalora": false,
+  "use_rslora": false
+}

systematicity_original_Swedish/Qwen3.5-4B-Base_systematicity_splits_original_features_train_systematicity_splits_original_features_test1/checkpoint-2190/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 %}

systematicity_original_Swedish/Qwen3.5-4B-Base_systematicity_splits_original_features_train_systematicity_splits_original_features_test1/checkpoint-2190/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|>"
+}

systematicity_original_Swedish/Qwen3.5-4B-Base_systematicity_splits_original_features_train_systematicity_splits_original_features_test1/checkpoint-2190/trainer_state.json

@@ -0,0 +1,519 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 5.0,
+  "eval_steps": 500,
+  "global_step": 2190,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "entropy": 1.4168990224599838,
+      "epoch": 0.11435105774728416,
+      "grad_norm": 1.3415299654006958,
+      "learning_rate": 4.424985043427545e-05,
+      "loss": 1.2160142517089845,
+      "mean_token_accuracy": 0.724469550549984,
+      "num_tokens": 96586.0,
+      "step": 50
+    },
+    {
+      "entropy": 0.7365697559714317,
+      "epoch": 0.22870211549456831,
+      "grad_norm": 0.9586130380630493,
+      "learning_rate": 8.940275904067897e-05,
+      "loss": 0.6445799255371094,
+      "mean_token_accuracy": 0.822787437736988,
+      "num_tokens": 190720.0,
+      "step": 100
+    },
+    {
+      "entropy": 0.691228820681572,
+      "epoch": 0.34305317324185247,
+      "grad_norm": 0.7875016331672668,
+      "learning_rate": 0.0001345556676470825,
+      "loss": 0.6011501312255859,
+      "mean_token_accuracy": 0.8325563898682594,
+      "num_tokens": 288515.0,
+      "step": 150
+    },
+    {
+      "entropy": 0.6706234656274319,
+      "epoch": 0.45740423098913663,
+      "grad_norm": 0.7867940068244934,
+      "learning_rate": 0.000179708576253486,
+      "loss": 0.5852980422973633,
+      "mean_token_accuracy": 0.83610415995121,
+      "num_tokens": 387874.0,
+      "step": 200
+    },
+    {
+      "entropy": 0.6558432649075985,
+      "epoch": 0.5717552887364208,
+      "grad_norm": 0.8049650192260742,
+      "learning_rate": 0.00022486148485988957,
+      "loss": 0.5714554977416992,
+      "mean_token_accuracy": 0.8380795314908027,
+      "num_tokens": 487865.0,
+      "step": 250
+    },
+    {
+      "entropy": 0.6612135827541351,
+      "epoch": 0.6861063464837049,
+      "grad_norm": 0.6776595711708069,
+      "learning_rate": 0.0002700143934662931,
+      "loss": 0.5815717315673828,
+      "mean_token_accuracy": 0.8377441626787185,
+      "num_tokens": 584269.0,
+      "step": 300
+    },
+    {
+      "entropy": 0.6668009972572326,
+      "epoch": 0.8004574042309891,
+      "grad_norm": 1.132367491722107,
+      "learning_rate": 0.0003151673020726966,
+      "loss": 0.5866946411132813,
+      "mean_token_accuracy": 0.8364784654974937,
+      "num_tokens": 676067.0,
+      "step": 350
+    },
+    {
+      "entropy": 0.6667388904094697,
+      "epoch": 0.9148084619782733,
+      "grad_norm": 0.9101859331130981,
+      "learning_rate": 0.0003603202106791001,
+      "loss": 0.5946685028076172,
+      "mean_token_accuracy": 0.8331978982686996,
+      "num_tokens": 774009.0,
+      "step": 400
+    },
+    {
+      "epoch": 1.0,
+      "eval_entropy": 0.5918735759536715,
+      "eval_loss": 0.6536311507225037,
+      "eval_mean_token_accuracy": 0.8276140002438622,
+      "eval_num_tokens": 847606.0,
+      "eval_runtime": 82.2575,
+      "eval_samples_per_second": 13.251,
+      "eval_steps_per_second": 1.666,
+      "step": 438
+    },
+    {
+      "entropy": 0.654371697890577,
+      "epoch": 1.0274442538593482,
+      "grad_norm": 0.923791229724884,
+      "learning_rate": 0.00039553187999846057,
+      "loss": 0.5879265594482422,
+      "mean_token_accuracy": 0.8356293612325252,
+      "num_tokens": 870214.0,
+      "step": 450
+    },
+    {
+      "entropy": 0.6167843751609325,
+      "epoch": 1.1417953116066324,
+      "grad_norm": 0.7943944334983826,
+      "learning_rate": 0.00039530582686207966,
+      "loss": 0.552466049194336,
+      "mean_token_accuracy": 0.8413123995065689,
+      "num_tokens": 971584.0,
+      "step": 500
+    },
+    {
+      "entropy": 0.6443017952144146,
+      "epoch": 1.2561463693539165,
+      "grad_norm": 1.018168568611145,
+      "learning_rate": 0.00039476615968757247,
+      "loss": 0.5773007583618164,
+      "mean_token_accuracy": 0.8377226310968399,
+      "num_tokens": 1063906.0,
+      "step": 550
+    },
+    {
+      "entropy": 0.6407185651361942,
+      "epoch": 1.3704974271012007,
+      "grad_norm": 0.9198986291885376,
+      "learning_rate": 0.00039391373526624385,
+      "loss": 0.572706184387207,
+      "mean_token_accuracy": 0.8367982304096222,
+      "num_tokens": 1159282.0,
+      "step": 600
+    },
+    {
+      "entropy": 0.6494526116549969,
+      "epoch": 1.4848484848484849,
+      "grad_norm": 1.0277528762817383,
+      "learning_rate": 0.00039274990693190134,
+      "loss": 0.578397102355957,
+      "mean_token_accuracy": 0.836915987432003,
+      "num_tokens": 1254998.0,
+      "step": 650
+    },
+    {
+      "entropy": 0.6396139825880528,
+      "epoch": 1.599199542595769,
+      "grad_norm": 0.8996608257293701,
+      "learning_rate": 0.0003912765224122631,
+      "loss": 0.5766008758544922,
+      "mean_token_accuracy": 0.8396538972854615,
+      "num_tokens": 1351877.0,
+      "step": 700
+    },
+    {
+      "entropy": 0.6245273813605309,
+      "epoch": 1.7135506003430532,
+      "grad_norm": 1.1999542713165283,
+      "learning_rate": 0.00038949592089545203,
+      "loss": 0.5718669509887695,
+      "mean_token_accuracy": 0.8406938418745995,
+      "num_tokens": 1448801.0,
+      "step": 750
+    },
+    {
+      "entropy": 0.6350119508802891,
+      "epoch": 1.8279016580903373,
+      "grad_norm": 0.9171736836433411,
+      "learning_rate": 0.00038741092931623213,
+      "loss": 0.5792041778564453,
+      "mean_token_accuracy": 0.8370065066218376,
+      "num_tokens": 1547040.0,
+      "step": 800
+    },
+    {
+      "entropy": 0.6106068694591522,
+      "epoch": 1.9422527158376215,
+      "grad_norm": 0.6869989037513733,
+      "learning_rate": 0.0003850248578678851,
+      "loss": 0.5748995590209961,
+      "mean_token_accuracy": 0.8383069199323654,
+      "num_tokens": 1646444.0,
+      "step": 850
+    },
+    {
+      "epoch": 2.0,
+      "eval_entropy": 0.5696834810893902,
+      "eval_loss": 0.6577129364013672,
+      "eval_mean_token_accuracy": 0.8307611911836332,
+      "eval_num_tokens": 1695212.0,
+      "eval_runtime": 81.951,
+      "eval_samples_per_second": 13.301,
+      "eval_steps_per_second": 1.672,
+      "step": 876
+    },
+    {
+      "entropy": 0.572038494088323,
+      "epoch": 2.0548885077186965,
+      "grad_norm": 0.7493197321891785,
+      "learning_rate": 0.00038234149474685104,
+      "loss": 0.5193290710449219,
+      "mean_token_accuracy": 0.8497188209277119,
+      "num_tokens": 1738782.0,
+      "step": 900
+    },
+    {
+      "entropy": 0.5478819286823273,
+      "epoch": 2.1692395654659804,
+      "grad_norm": 1.2549943923950195,
+      "learning_rate": 0.0003793651001384773,
+      "loss": 0.4793633270263672,
+      "mean_token_accuracy": 0.8575673401355743,
+      "num_tokens": 1837910.0,
+      "step": 950
+    },
+    {
+      "entropy": 0.5431128491461277,
+      "epoch": 2.283590623213265,
+      "grad_norm": 0.8713223338127136,
+      "learning_rate": 0.00037610039945342474,
+      "loss": 0.4849379348754883,
+      "mean_token_accuracy": 0.8562448808550834,
+      "num_tokens": 1937110.0,
+      "step": 1000
+    },
+    {
+      "entropy": 0.5498340383172036,
+      "epoch": 2.3979416809605487,
+      "grad_norm": 0.6592629551887512,
+      "learning_rate": 0.00037255257582546804,
+      "loss": 0.493680419921875,
+      "mean_token_accuracy": 0.854153895676136,
+      "num_tokens": 2036890.0,
+      "step": 1050
+    },
+    {
+      "entropy": 0.5554873237013817,
+      "epoch": 2.512292738707833,
+      "grad_norm": 0.7878748774528503,
+      "learning_rate": 0.0003687272618826019,
+      "loss": 0.4951488494873047,
+      "mean_token_accuracy": 0.8540353265404701,
+      "num_tokens": 2137239.0,
+      "step": 1100
+    },
+    {
+      "entropy": 0.552360344529152,
+      "epoch": 2.6266437964551175,
+      "grad_norm": 0.7380415797233582,
+      "learning_rate": 0.00036463053080451683,
+      "loss": 0.49989810943603513,
+      "mean_token_accuracy": 0.8533406674861908,
+      "num_tokens": 2231081.0,
+      "step": 1150
+    },
+    {
+      "entropy": 0.5345852719247342,
+      "epoch": 2.7409948542024014,
+      "grad_norm": 0.7600613832473755,
+      "learning_rate": 0.0003602688866806424,
+      "loss": 0.5011516189575196,
+      "mean_token_accuracy": 0.8527707865834236,
+      "num_tokens": 2325503.0,
+      "step": 1200
+    },
+    {
+      "entropy": 0.532319468408823,
+      "epoch": 2.8553459119496853,
+      "grad_norm": 0.8197540640830994,
+      "learning_rate": 0.000355649254184066,
+      "loss": 0.5023023605346679,
+      "mean_token_accuracy": 0.8530397671461105,
+      "num_tokens": 2420627.0,
+      "step": 1250
+    },
+    {
+      "entropy": 0.5198138700425625,
+      "epoch": 2.9696969696969697,
+      "grad_norm": 0.8660516142845154,
+      "learning_rate": 0.00035077896757771987,
+      "loss": 0.4941487121582031,
+      "mean_token_accuracy": 0.855004439651966,
+      "num_tokens": 2518051.0,
+      "step": 1300
+    },
+    {
+      "epoch": 3.0,
+      "eval_entropy": 0.5642598084289662,
+      "eval_loss": 0.6125267744064331,
+      "eval_mean_token_accuracy": 0.8375998204641969,
+      "eval_num_tokens": 2542818.0,
+      "eval_runtime": 81.5524,
+      "eval_samples_per_second": 13.366,
+      "eval_steps_per_second": 1.68,
+      "step": 1314
+    },
+    {
+      "entropy": 0.45835615868495805,
+      "epoch": 3.0823327615780447,
+      "grad_norm": 0.734519362449646,
+      "learning_rate": 0.00034566575907029245,
+      "loss": 0.4334117889404297,
+      "mean_token_accuracy": 0.8698135275526095,
+      "num_tokens": 2613861.0,
+      "step": 1350
+    },
+    {
+      "entropy": 0.44648877054452896,
+      "epoch": 3.1966838193253286,
+      "grad_norm": 0.687981128692627,
+      "learning_rate": 0.00034031774654034867,
+      "loss": 0.40152584075927733,
+      "mean_token_accuracy": 0.8759277358651161,
+      "num_tokens": 2709451.0,
+      "step": 1400
+    },
+    {
+      "entropy": 0.4593352344632149,
+      "epoch": 3.311034877072613,
+      "grad_norm": 0.6310222148895264,
+      "learning_rate": 0.0003347434206481497,
+      "loss": 0.41885005950927734,
+      "mean_token_accuracy": 0.8714999267458916,
+      "num_tokens": 2801336.0,
+      "step": 1450
+    },
+    {
+      "entropy": 0.46372457168996334,
+      "epoch": 3.425385934819897,
+      "grad_norm": 0.6795068979263306,
+      "learning_rate": 0.0003289516313556339,
+      "loss": 0.4194014358520508,
+      "mean_token_accuracy": 0.8709167090058326,
+      "num_tokens": 2898945.0,
+      "step": 1500
+    },
+    {
+      "entropy": 0.4548304344713688,
+      "epoch": 3.5397369925671813,
+      "grad_norm": 0.9671820402145386,
+      "learning_rate": 0.00032295157387595997,
+      "loss": 0.4178035736083984,
+      "mean_token_accuracy": 0.8710319435596466,
+      "num_tokens": 2993378.0,
+      "step": 1550
+    },
+    {
+      "entropy": 0.4532127542793751,
+      "epoch": 3.6540880503144653,
+      "grad_norm": 0.7656959295272827,
+      "learning_rate": 0.0003167527740749187,
+      "loss": 0.42567527770996094,
+      "mean_token_accuracy": 0.869033499956131,
+      "num_tokens": 3091323.0,
+      "step": 1600
+    },
+    {
+      "entropy": 0.4460975781828165,
+      "epoch": 3.7684391080617496,
+      "grad_norm": 0.7811372876167297,
+      "learning_rate": 0.0003103650733473916,
+      "loss": 0.4148756790161133,
+      "mean_token_accuracy": 0.8709280017018318,
+      "num_tokens": 3193194.0,
+      "step": 1650
+    },
+    {
+      "entropy": 0.47125090911984446,
+      "epoch": 3.8827901658090336,
+      "grad_norm": 0.8664485216140747,
+      "learning_rate": 0.00030379861299286587,
+      "loss": 0.4209564590454102,
+      "mean_token_accuracy": 0.8695067319273949,
+      "num_tokens": 3292840.0,
+      "step": 1700
+    },
+    {
+      "entropy": 0.4641740974783897,
+      "epoch": 3.997141223556318,
+      "grad_norm": 0.85616534948349,
+      "learning_rate": 0.00029706381811481263,
+      "loss": 0.4319478988647461,
+      "mean_token_accuracy": 0.8689105224609375,
+      "num_tokens": 3387511.0,
+      "step": 1750
+    },
+    {
+      "epoch": 4.0,
+      "eval_entropy": 0.4623802898138979,
+      "eval_loss": 0.6707634925842285,
+      "eval_mean_token_accuracy": 0.8327476525828786,
+      "eval_num_tokens": 3390424.0,
+      "eval_runtime": 81.9106,
+      "eval_samples_per_second": 13.307,
+      "eval_steps_per_second": 1.673,
+      "step": 1752
+    },
+    {
+      "entropy": 0.33750920285125674,
+      "epoch": 4.109777015437393,
+      "grad_norm": 0.6137083172798157,
+      "learning_rate": 0.00029017138106948915,
+      "loss": 0.31130544662475584,
+      "mean_token_accuracy": 0.8997744660692166,
+      "num_tokens": 3483795.0,
+      "step": 1800
+    },
+    {
+      "entropy": 0.3504758469760418,
+      "epoch": 4.224128073184677,
+      "grad_norm": 0.6796069741249084,
+      "learning_rate": 0.00028313224449044337,
+      "loss": 0.31956762313842774,
+      "mean_token_accuracy": 0.895492787361145,
+      "num_tokens": 3576928.0,
+      "step": 1850
+    },
+    {
+      "entropy": 0.34297160781919955,
+      "epoch": 4.338479130931961,
+      "grad_norm": 0.7507141828536987,
+      "learning_rate": 0.00027595758391566995,
+      "loss": 0.31722068786621094,
+      "mean_token_accuracy": 0.8957494455575943,
+      "num_tokens": 3674125.0,
+      "step": 1900
+    },
+    {
+      "entropy": 0.3572557730227709,
+      "epoch": 4.452830188679245,
+      "grad_norm": 0.7509586215019226,
+      "learning_rate": 0.00026865879004500166,
+      "loss": 0.32728019714355466,
+      "mean_token_accuracy": 0.894504280090332,
+      "num_tokens": 3769853.0,
+      "step": 1950
+    },
+    {
+      "entropy": 0.35588217943906786,
+      "epoch": 4.56718124642653,
+      "grad_norm": 0.6168456077575684,
+      "learning_rate": 0.0002612474506559027,
+      "loss": 0.3304343795776367,
+      "mean_token_accuracy": 0.8935999310016632,
+      "num_tokens": 3867574.0,
+      "step": 2000
+    },
+    {
+      "entropy": 0.3554471937566996,
+      "epoch": 4.681532304173814,
+      "grad_norm": 0.6922969222068787,
+      "learning_rate": 0.00025373533220637623,
+      "loss": 0.3315126037597656,
+      "mean_token_accuracy": 0.8925720381736756,
+      "num_tokens": 3969484.0,
+      "step": 2050
+    },
+    {
+      "entropy": 0.35552985966205597,
+      "epoch": 4.795883361921097,
+      "grad_norm": 0.6013802289962769,
+      "learning_rate": 0.00024613436115419326,
+      "loss": 0.3315456771850586,
+      "mean_token_accuracy": 0.8929180511832238,
+      "num_tokens": 4063480.0,
+      "step": 2100
+    },
+    {
+      "entropy": 0.3478537338972092,
+      "epoch": 4.910234419668382,
+      "grad_norm": 0.8090653419494629,
+      "learning_rate": 0.00023845660502210138,
+      "loss": 0.32579368591308594,
+      "mean_token_accuracy": 0.893961855173111,
+      "num_tokens": 4162925.0,
+      "step": 2150
+    },
+    {
+      "epoch": 5.0,
+      "eval_entropy": 0.38475005339532004,
+      "eval_loss": 0.6992377042770386,
+      "eval_mean_token_accuracy": 0.8384414993933518,
+      "eval_num_tokens": 4238030.0,
+      "eval_runtime": 81.1991,
+      "eval_samples_per_second": 13.424,
+      "eval_steps_per_second": 1.687,
+      "step": 2190
+    }
+  ],
+  "logging_steps": 50,
+  "max_steps": 4380,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 10,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 1.758142122273116e+17,
+  "train_batch_size": 4,
+  "trial_name": null,
+  "trial_params": null
+}

systematicity_original_Swedish/Qwen3.5-4B-Base_systematicity_splits_original_features_train_systematicity_splits_original_features_test1/checkpoint-2628/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

systematicity_original_Swedish/Qwen3.5-4B-Base_systematicity_splits_original_features_train_systematicity_splits_original_features_test1/checkpoint-2628/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": 0.026477173062682283,
+  "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": [
+    "q_proj",
+    "k_proj",
+    "down_proj",
+    "v_proj",
+    "gate_proj",
+    "o_proj",
+    "up_proj"
+  ],
+  "target_parameters": null,
+  "task_type": "CAUSAL_LM",
+  "trainable_token_indices": null,
+  "use_dora": false,
+  "use_qalora": false,
+  "use_rslora": false
+}

systematicity_original_Swedish/Qwen3.5-4B-Base_systematicity_splits_original_features_train_systematicity_splits_original_features_test1/checkpoint-2628/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 %}

systematicity_original_Swedish/Qwen3.5-4B-Base_systematicity_splits_original_features_train_systematicity_splits_original_features_test1/checkpoint-2628/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|>"
+}

systematicity_original_Swedish/Qwen3.5-4B-Base_systematicity_splits_original_features_train_systematicity_splits_original_features_test1/checkpoint-2628/trainer_state.json

@@ -0,0 +1,620 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 6.0,
+  "eval_steps": 500,
+  "global_step": 2628,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "entropy": 1.4168990224599838,
+      "epoch": 0.11435105774728416,
+      "grad_norm": 1.3415299654006958,
+      "learning_rate": 4.424985043427545e-05,
+      "loss": 1.2160142517089845,
+      "mean_token_accuracy": 0.724469550549984,
+      "num_tokens": 96586.0,
+      "step": 50
+    },
+    {
+      "entropy": 0.7365697559714317,
+      "epoch": 0.22870211549456831,
+      "grad_norm": 0.9586130380630493,
+      "learning_rate": 8.940275904067897e-05,
+      "loss": 0.6445799255371094,
+      "mean_token_accuracy": 0.822787437736988,
+      "num_tokens": 190720.0,
+      "step": 100
+    },
+    {
+      "entropy": 0.691228820681572,
+      "epoch": 0.34305317324185247,
+      "grad_norm": 0.7875016331672668,
+      "learning_rate": 0.0001345556676470825,
+      "loss": 0.6011501312255859,
+      "mean_token_accuracy": 0.8325563898682594,
+      "num_tokens": 288515.0,
+      "step": 150
+    },
+    {
+      "entropy": 0.6706234656274319,
+      "epoch": 0.45740423098913663,
+      "grad_norm": 0.7867940068244934,
+      "learning_rate": 0.000179708576253486,
+      "loss": 0.5852980422973633,
+      "mean_token_accuracy": 0.83610415995121,
+      "num_tokens": 387874.0,
+      "step": 200
+    },
+    {
+      "entropy": 0.6558432649075985,
+      "epoch": 0.5717552887364208,
+      "grad_norm": 0.8049650192260742,
+      "learning_rate": 0.00022486148485988957,
+      "loss": 0.5714554977416992,
+      "mean_token_accuracy": 0.8380795314908027,
+      "num_tokens": 487865.0,
+      "step": 250
+    },
+    {
+      "entropy": 0.6612135827541351,
+      "epoch": 0.6861063464837049,
+      "grad_norm": 0.6776595711708069,
+      "learning_rate": 0.0002700143934662931,
+      "loss": 0.5815717315673828,
+      "mean_token_accuracy": 0.8377441626787185,
+      "num_tokens": 584269.0,
+      "step": 300
+    },
+    {
+      "entropy": 0.6668009972572326,
+      "epoch": 0.8004574042309891,
+      "grad_norm": 1.132367491722107,
+      "learning_rate": 0.0003151673020726966,
+      "loss": 0.5866946411132813,
+      "mean_token_accuracy": 0.8364784654974937,
+      "num_tokens": 676067.0,
+      "step": 350
+    },
+    {
+      "entropy": 0.6667388904094697,
+      "epoch": 0.9148084619782733,
+      "grad_norm": 0.9101859331130981,
+      "learning_rate": 0.0003603202106791001,
+      "loss": 0.5946685028076172,
+      "mean_token_accuracy": 0.8331978982686996,
+      "num_tokens": 774009.0,
+      "step": 400
+    },
+    {
+      "epoch": 1.0,
+      "eval_entropy": 0.5918735759536715,
+      "eval_loss": 0.6536311507225037,
+      "eval_mean_token_accuracy": 0.8276140002438622,
+      "eval_num_tokens": 847606.0,
+      "eval_runtime": 82.2575,
+      "eval_samples_per_second": 13.251,
+      "eval_steps_per_second": 1.666,
+      "step": 438
+    },
+    {
+      "entropy": 0.654371697890577,
+      "epoch": 1.0274442538593482,
+      "grad_norm": 0.923791229724884,
+      "learning_rate": 0.00039553187999846057,
+      "loss": 0.5879265594482422,
+      "mean_token_accuracy": 0.8356293612325252,
+      "num_tokens": 870214.0,
+      "step": 450
+    },
+    {
+      "entropy": 0.6167843751609325,
+      "epoch": 1.1417953116066324,
+      "grad_norm": 0.7943944334983826,
+      "learning_rate": 0.00039530582686207966,
+      "loss": 0.552466049194336,
+      "mean_token_accuracy": 0.8413123995065689,
+      "num_tokens": 971584.0,
+      "step": 500
+    },
+    {
+      "entropy": 0.6443017952144146,
+      "epoch": 1.2561463693539165,
+      "grad_norm": 1.018168568611145,
+      "learning_rate": 0.00039476615968757247,
+      "loss": 0.5773007583618164,
+      "mean_token_accuracy": 0.8377226310968399,
+      "num_tokens": 1063906.0,
+      "step": 550
+    },
+    {
+      "entropy": 0.6407185651361942,
+      "epoch": 1.3704974271012007,
+      "grad_norm": 0.9198986291885376,
+      "learning_rate": 0.00039391373526624385,
+      "loss": 0.572706184387207,
+      "mean_token_accuracy": 0.8367982304096222,
+      "num_tokens": 1159282.0,
+      "step": 600
+    },
+    {
+      "entropy": 0.6494526116549969,
+      "epoch": 1.4848484848484849,
+      "grad_norm": 1.0277528762817383,
+      "learning_rate": 0.00039274990693190134,
+      "loss": 0.578397102355957,
+      "mean_token_accuracy": 0.836915987432003,
+      "num_tokens": 1254998.0,
+      "step": 650
+    },
+    {
+      "entropy": 0.6396139825880528,
+      "epoch": 1.599199542595769,
+      "grad_norm": 0.8996608257293701,
+      "learning_rate": 0.0003912765224122631,
+      "loss": 0.5766008758544922,
+      "mean_token_accuracy": 0.8396538972854615,
+      "num_tokens": 1351877.0,
+      "step": 700
+    },
+    {
+      "entropy": 0.6245273813605309,
+      "epoch": 1.7135506003430532,
+      "grad_norm": 1.1999542713165283,
+      "learning_rate": 0.00038949592089545203,
+      "loss": 0.5718669509887695,
+      "mean_token_accuracy": 0.8406938418745995,
+      "num_tokens": 1448801.0,
+      "step": 750
+    },
+    {
+      "entropy": 0.6350119508802891,
+      "epoch": 1.8279016580903373,
+      "grad_norm": 0.9171736836433411,
+      "learning_rate": 0.00038741092931623213,
+      "loss": 0.5792041778564453,
+      "mean_token_accuracy": 0.8370065066218376,
+      "num_tokens": 1547040.0,
+      "step": 800
+    },
+    {
+      "entropy": 0.6106068694591522,
+      "epoch": 1.9422527158376215,
+      "grad_norm": 0.6869989037513733,
+      "learning_rate": 0.0003850248578678851,
+      "loss": 0.5748995590209961,
+      "mean_token_accuracy": 0.8383069199323654,
+      "num_tokens": 1646444.0,
+      "step": 850
+    },
+    {
+      "epoch": 2.0,
+      "eval_entropy": 0.5696834810893902,
+      "eval_loss": 0.6577129364013672,
+      "eval_mean_token_accuracy": 0.8307611911836332,
+      "eval_num_tokens": 1695212.0,
+      "eval_runtime": 81.951,
+      "eval_samples_per_second": 13.301,
+      "eval_steps_per_second": 1.672,
+      "step": 876
+    },
+    {
+      "entropy": 0.572038494088323,
+      "epoch": 2.0548885077186965,
+      "grad_norm": 0.7493197321891785,
+      "learning_rate": 0.00038234149474685104,
+      "loss": 0.5193290710449219,
+      "mean_token_accuracy": 0.8497188209277119,
+      "num_tokens": 1738782.0,
+      "step": 900
+    },
+    {
+      "entropy": 0.5478819286823273,
+      "epoch": 2.1692395654659804,
+      "grad_norm": 1.2549943923950195,
+      "learning_rate": 0.0003793651001384773,
+      "loss": 0.4793633270263672,
+      "mean_token_accuracy": 0.8575673401355743,
+      "num_tokens": 1837910.0,
+      "step": 950
+    },
+    {
+      "entropy": 0.5431128491461277,
+      "epoch": 2.283590623213265,
+      "grad_norm": 0.8713223338127136,
+      "learning_rate": 0.00037610039945342474,
+      "loss": 0.4849379348754883,
+      "mean_token_accuracy": 0.8562448808550834,
+      "num_tokens": 1937110.0,
+      "step": 1000
+    },
+    {
+      "entropy": 0.5498340383172036,
+      "epoch": 2.3979416809605487,
+      "grad_norm": 0.6592629551887512,
+      "learning_rate": 0.00037255257582546804,
+      "loss": 0.493680419921875,
+      "mean_token_accuracy": 0.854153895676136,
+      "num_tokens": 2036890.0,
+      "step": 1050
+    },
+    {
+      "entropy": 0.5554873237013817,
+      "epoch": 2.512292738707833,
+      "grad_norm": 0.7878748774528503,
+      "learning_rate": 0.0003687272618826019,
+      "loss": 0.4951488494873047,
+      "mean_token_accuracy": 0.8540353265404701,
+      "num_tokens": 2137239.0,
+      "step": 1100
+    },
+    {
+      "entropy": 0.552360344529152,
+      "epoch": 2.6266437964551175,
+      "grad_norm": 0.7380415797233582,
+      "learning_rate": 0.00036463053080451683,
+      "loss": 0.49989810943603513,
+      "mean_token_accuracy": 0.8533406674861908,
+      "num_tokens": 2231081.0,
+      "step": 1150
+    },
+    {
+      "entropy": 0.5345852719247342,
+      "epoch": 2.7409948542024014,
+      "grad_norm": 0.7600613832473755,
+      "learning_rate": 0.0003602688866806424,
+      "loss": 0.5011516189575196,
+      "mean_token_accuracy": 0.8527707865834236,
+      "num_tokens": 2325503.0,
+      "step": 1200
+    },
+    {
+      "entropy": 0.532319468408823,
+      "epoch": 2.8553459119496853,
+      "grad_norm": 0.8197540640830994,
+      "learning_rate": 0.000355649254184066,
+      "loss": 0.5023023605346679,
+      "mean_token_accuracy": 0.8530397671461105,
+      "num_tokens": 2420627.0,
+      "step": 1250
+    },
+    {
+      "entropy": 0.5198138700425625,
+      "epoch": 2.9696969696969697,
+      "grad_norm": 0.8660516142845154,
+      "learning_rate": 0.00035077896757771987,
+      "loss": 0.4941487121582031,
+      "mean_token_accuracy": 0.855004439651966,
+      "num_tokens": 2518051.0,
+      "step": 1300
+    },
+    {
+      "epoch": 3.0,
+      "eval_entropy": 0.5642598084289662,
+      "eval_loss": 0.6125267744064331,
+      "eval_mean_token_accuracy": 0.8375998204641969,
+      "eval_num_tokens": 2542818.0,
+      "eval_runtime": 81.5524,
+      "eval_samples_per_second": 13.366,
+      "eval_steps_per_second": 1.68,
+      "step": 1314
+    },
+    {
+      "entropy": 0.45835615868495805,
+      "epoch": 3.0823327615780447,
+      "grad_norm": 0.734519362449646,
+      "learning_rate": 0.00034566575907029245,
+      "loss": 0.4334117889404297,
+      "mean_token_accuracy": 0.8698135275526095,
+      "num_tokens": 2613861.0,
+      "step": 1350
+    },
+    {
+      "entropy": 0.44648877054452896,
+      "epoch": 3.1966838193253286,
+      "grad_norm": 0.687981128692627,
+      "learning_rate": 0.00034031774654034867,
+      "loss": 0.40152584075927733,
+      "mean_token_accuracy": 0.8759277358651161,
+      "num_tokens": 2709451.0,
+      "step": 1400
+    },
+    {
+      "entropy": 0.4593352344632149,
+      "epoch": 3.311034877072613,
+      "grad_norm": 0.6310222148895264,
+      "learning_rate": 0.0003347434206481497,
+      "loss": 0.41885005950927734,
+      "mean_token_accuracy": 0.8714999267458916,
+      "num_tokens": 2801336.0,
+      "step": 1450
+    },
+    {
+      "entropy": 0.46372457168996334,
+      "epoch": 3.425385934819897,
+      "grad_norm": 0.6795068979263306,
+      "learning_rate": 0.0003289516313556339,
+      "loss": 0.4194014358520508,
+      "mean_token_accuracy": 0.8709167090058326,
+      "num_tokens": 2898945.0,
+      "step": 1500
+    },
+    {
+      "entropy": 0.4548304344713688,
+      "epoch": 3.5397369925671813,
+      "grad_norm": 0.9671820402145386,
+      "learning_rate": 0.00032295157387595997,
+      "loss": 0.4178035736083984,
+      "mean_token_accuracy": 0.8710319435596466,
+      "num_tokens": 2993378.0,
+      "step": 1550
+    },
+    {
+      "entropy": 0.4532127542793751,
+      "epoch": 3.6540880503144653,
+      "grad_norm": 0.7656959295272827,
+      "learning_rate": 0.0003167527740749187,
+      "loss": 0.42567527770996094,
+      "mean_token_accuracy": 0.869033499956131,
+      "num_tokens": 3091323.0,
+      "step": 1600
+    },
+    {
+      "entropy": 0.4460975781828165,
+      "epoch": 3.7684391080617496,
+      "grad_norm": 0.7811372876167297,
+      "learning_rate": 0.0003103650733473916,
+      "loss": 0.4148756790161133,
+      "mean_token_accuracy": 0.8709280017018318,
+      "num_tokens": 3193194.0,
+      "step": 1650
+    },
+    {
+      "entropy": 0.47125090911984446,
+      "epoch": 3.8827901658090336,
+      "grad_norm": 0.8664485216140747,
+      "learning_rate": 0.00030379861299286587,
+      "loss": 0.4209564590454102,
+      "mean_token_accuracy": 0.8695067319273949,
+      "num_tokens": 3292840.0,
+      "step": 1700
+    },
+    {
+      "entropy": 0.4641740974783897,
+      "epoch": 3.997141223556318,
+      "grad_norm": 0.85616534948349,
+      "learning_rate": 0.00029706381811481263,
+      "loss": 0.4319478988647461,
+      "mean_token_accuracy": 0.8689105224609375,
+      "num_tokens": 3387511.0,
+      "step": 1750
+    },
+    {
+      "epoch": 4.0,
+      "eval_entropy": 0.4623802898138979,
+      "eval_loss": 0.6707634925842285,
+      "eval_mean_token_accuracy": 0.8327476525828786,
+      "eval_num_tokens": 3390424.0,
+      "eval_runtime": 81.9106,
+      "eval_samples_per_second": 13.307,
+      "eval_steps_per_second": 1.673,
+      "step": 1752
+    },
+    {
+      "entropy": 0.33750920285125674,
+      "epoch": 4.109777015437393,
+      "grad_norm": 0.6137083172798157,
+      "learning_rate": 0.00029017138106948915,
+      "loss": 0.31130544662475584,
+      "mean_token_accuracy": 0.8997744660692166,
+      "num_tokens": 3483795.0,
+      "step": 1800
+    },
+    {
+      "entropy": 0.3504758469760418,
+      "epoch": 4.224128073184677,
+      "grad_norm": 0.6796069741249084,
+      "learning_rate": 0.00028313224449044337,
+      "loss": 0.31956762313842774,
+      "mean_token_accuracy": 0.895492787361145,
+      "num_tokens": 3576928.0,
+      "step": 1850
+    },
+    {
+      "entropy": 0.34297160781919955,
+      "epoch": 4.338479130931961,
+      "grad_norm": 0.7507141828536987,
+      "learning_rate": 0.00027595758391566995,
+      "loss": 0.31722068786621094,
+      "mean_token_accuracy": 0.8957494455575943,
+      "num_tokens": 3674125.0,
+      "step": 1900
+    },
+    {
+      "entropy": 0.3572557730227709,
+      "epoch": 4.452830188679245,
+      "grad_norm": 0.7509586215019226,
+      "learning_rate": 0.00026865879004500166,
+      "loss": 0.32728019714355466,
+      "mean_token_accuracy": 0.894504280090332,
+      "num_tokens": 3769853.0,
+      "step": 1950
+    },
+    {
+      "entropy": 0.35588217943906786,
+      "epoch": 4.56718124642653,
+      "grad_norm": 0.6168456077575684,
+      "learning_rate": 0.0002612474506559027,
+      "loss": 0.3304343795776367,
+      "mean_token_accuracy": 0.8935999310016632,
+      "num_tokens": 3867574.0,
+      "step": 2000
+    },
+    {
+      "entropy": 0.3554471937566996,
+      "epoch": 4.681532304173814,
+      "grad_norm": 0.6922969222068787,
+      "learning_rate": 0.00025373533220637623,
+      "loss": 0.3315126037597656,
+      "mean_token_accuracy": 0.8925720381736756,
+      "num_tokens": 3969484.0,
+      "step": 2050
+    },
+    {
+      "entropy": 0.35552985966205597,
+      "epoch": 4.795883361921097,
+      "grad_norm": 0.6013802289962769,
+      "learning_rate": 0.00024613436115419326,
+      "loss": 0.3315456771850586,
+      "mean_token_accuracy": 0.8929180511832238,
+      "num_tokens": 4063480.0,
+      "step": 2100
+    },
+    {
+      "entropy": 0.3478537338972092,
+      "epoch": 4.910234419668382,
+      "grad_norm": 0.8090653419494629,
+      "learning_rate": 0.00023845660502210138,
+      "loss": 0.32579368591308594,
+      "mean_token_accuracy": 0.893961855173111,
+      "num_tokens": 4162925.0,
+      "step": 2150
+    },
+    {
+      "epoch": 5.0,
+      "eval_entropy": 0.38475005339532004,
+      "eval_loss": 0.6992377042770386,
+      "eval_mean_token_accuracy": 0.8384414993933518,
+      "eval_num_tokens": 4238030.0,
+      "eval_runtime": 81.1991,
+      "eval_samples_per_second": 13.424,
+      "eval_steps_per_second": 1.687,
+      "step": 2190
+    },
+    {
+      "entropy": 0.3381291570397198,
+      "epoch": 5.022870211549457,
+      "grad_norm": 0.7052969932556152,
+      "learning_rate": 0.00023071425323907478,
+      "loss": 0.3124395370483398,
+      "mean_token_accuracy": 0.8997272504162668,
+      "num_tokens": 4257610.0,
+      "step": 2200
+    },
+    {
+      "entropy": 0.2494013351202011,
+      "epoch": 5.137221269296741,
+      "grad_norm": 0.8653550148010254,
+      "learning_rate": 0.0002229195977880219,
+      "loss": 0.21398515701293946,
+      "mean_token_accuracy": 0.9277590665221215,
+      "num_tokens": 4353703.0,
+      "step": 2250
+    },
+    {
+      "entropy": 0.27507017262279987,
+      "epoch": 5.251572327044025,
+      "grad_norm": 0.7887904644012451,
+      "learning_rate": 0.00021508501369067626,
+      "loss": 0.22058496475219727,
+      "mean_token_accuracy": 0.9253441894054413,
+      "num_tokens": 4449829.0,
+      "step": 2300
+    },
+    {
+      "entropy": 0.2582047886401415,
+      "epoch": 5.3659233847913095,
+      "grad_norm": 0.8728304505348206,
+      "learning_rate": 0.0002072229393606514,
+      "loss": 0.22437936782836915,
+      "mean_token_accuracy": 0.9235965391993523,
+      "num_tokens": 4545644.0,
+      "step": 2350
+    },
+    {
+      "entropy": 0.26232284381985665,
+      "epoch": 5.480274442538594,
+      "grad_norm": 0.6688716411590576,
+      "learning_rate": 0.00019934585685585428,
+      "loss": 0.23260833740234374,
+      "mean_token_accuracy": 0.9217587423324585,
+      "num_tokens": 4637941.0,
+      "step": 2400
+    },
+    {
+      "entropy": 0.2619731902331114,
+      "epoch": 5.594625500285877,
+      "grad_norm": 0.597118616104126,
+      "learning_rate": 0.00019146627206160645,
+      "loss": 0.22075172424316405,
+      "mean_token_accuracy": 0.9252614346146584,
+      "num_tokens": 4738175.0,
+      "step": 2450
+    },
+    {
+      "entropy": 0.26591138120740654,
+      "epoch": 5.708976558033162,
+      "grad_norm": 0.7602102160453796,
+      "learning_rate": 0.00018359669483593703,
+      "loss": 0.22200309753417968,
+      "mean_token_accuracy": 0.9250017151236534,
+      "num_tokens": 4837766.0,
+      "step": 2500
+    },
+    {
+      "entropy": 0.2587306099012494,
+      "epoch": 5.823327615780446,
+      "grad_norm": 0.6875442862510681,
+      "learning_rate": 0.00017574961914856833,
+      "loss": 0.22627996444702148,
+      "mean_token_accuracy": 0.9234379661083222,
+      "num_tokens": 4935670.0,
+      "step": 2550
+    },
+    {
+      "entropy": 0.2689222282171249,
+      "epoch": 5.9376786735277305,
+      "grad_norm": 0.654289186000824,
+      "learning_rate": 0.00016793750324512556,
+      "loss": 0.23186542510986327,
+      "mean_token_accuracy": 0.9214966455101967,
+      "num_tokens": 5030799.0,
+      "step": 2600
+    },
+    {
+      "epoch": 6.0,
+      "eval_entropy": 0.34485617432281046,
+      "eval_loss": 0.7253938317298889,
+      "eval_mean_token_accuracy": 0.8377144193127207,
+      "eval_num_tokens": 5085636.0,
+      "eval_runtime": 81.5281,
+      "eval_samples_per_second": 13.37,
+      "eval_steps_per_second": 1.68,
+      "step": 2628
+    }
+  ],
+  "logging_steps": 50,
+  "max_steps": 4380,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 10,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 2.1097379588658586e+17,
+  "train_batch_size": 4,
+  "trial_name": null,
+  "trial_params": null
+}

systematicity_original_Swedish/Qwen3.5-4B-Base_systematicity_splits_original_features_train_systematicity_splits_original_features_test1/checkpoint-3066/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

systematicity_original_Swedish/Qwen3.5-4B-Base_systematicity_splits_original_features_train_systematicity_splits_original_features_test1/checkpoint-3066/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": 0.026477173062682283,
+  "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": [
+    "q_proj",
+    "k_proj",
+    "down_proj",
+    "v_proj",
+    "gate_proj",
+    "o_proj",
+    "up_proj"
+  ],
+  "target_parameters": null,
+  "task_type": "CAUSAL_LM",
+  "trainable_token_indices": null,
+  "use_dora": false,
+  "use_qalora": false,
+  "use_rslora": false
+}

systematicity_original_Swedish/Qwen3.5-4B-Base_systematicity_splits_original_features_train_systematicity_splits_original_features_test1/checkpoint-3066/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 %}

systematicity_original_Swedish/Qwen3.5-4B-Base_systematicity_splits_original_features_train_systematicity_splits_original_features_test1/checkpoint-3066/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|>"
+}

systematicity_original_Swedish/Qwen3.5-4B-Base_systematicity_splits_original_features_train_systematicity_splits_original_features_test1/checkpoint-3066/trainer_state.json

@@ -0,0 +1,721 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 7.0,
+  "eval_steps": 500,
+  "global_step": 3066,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "entropy": 1.4168990224599838,
+      "epoch": 0.11435105774728416,
+      "grad_norm": 1.3415299654006958,
+      "learning_rate": 4.424985043427545e-05,
+      "loss": 1.2160142517089845,
+      "mean_token_accuracy": 0.724469550549984,
+      "num_tokens": 96586.0,
+      "step": 50
+    },
+    {
+      "entropy": 0.7365697559714317,
+      "epoch": 0.22870211549456831,
+      "grad_norm": 0.9586130380630493,
+      "learning_rate": 8.940275904067897e-05,
+      "loss": 0.6445799255371094,
+      "mean_token_accuracy": 0.822787437736988,
+      "num_tokens": 190720.0,
+      "step": 100
+    },
+    {
+      "entropy": 0.691228820681572,
+      "epoch": 0.34305317324185247,
+      "grad_norm": 0.7875016331672668,
+      "learning_rate": 0.0001345556676470825,
+      "loss": 0.6011501312255859,
+      "mean_token_accuracy": 0.8325563898682594,
+      "num_tokens": 288515.0,
+      "step": 150
+    },
+    {
+      "entropy": 0.6706234656274319,
+      "epoch": 0.45740423098913663,
+      "grad_norm": 0.7867940068244934,
+      "learning_rate": 0.000179708576253486,
+      "loss": 0.5852980422973633,
+      "mean_token_accuracy": 0.83610415995121,
+      "num_tokens": 387874.0,
+      "step": 200
+    },
+    {
+      "entropy": 0.6558432649075985,
+      "epoch": 0.5717552887364208,
+      "grad_norm": 0.8049650192260742,
+      "learning_rate": 0.00022486148485988957,
+      "loss": 0.5714554977416992,
+      "mean_token_accuracy": 0.8380795314908027,
+      "num_tokens": 487865.0,
+      "step": 250
+    },
+    {
+      "entropy": 0.6612135827541351,
+      "epoch": 0.6861063464837049,
+      "grad_norm": 0.6776595711708069,
+      "learning_rate": 0.0002700143934662931,
+      "loss": 0.5815717315673828,
+      "mean_token_accuracy": 0.8377441626787185,
+      "num_tokens": 584269.0,
+      "step": 300
+    },
+    {
+      "entropy": 0.6668009972572326,
+      "epoch": 0.8004574042309891,
+      "grad_norm": 1.132367491722107,
+      "learning_rate": 0.0003151673020726966,
+      "loss": 0.5866946411132813,
+      "mean_token_accuracy": 0.8364784654974937,
+      "num_tokens": 676067.0,
+      "step": 350
+    },
+    {
+      "entropy": 0.6667388904094697,
+      "epoch": 0.9148084619782733,
+      "grad_norm": 0.9101859331130981,
+      "learning_rate": 0.0003603202106791001,
+      "loss": 0.5946685028076172,
+      "mean_token_accuracy": 0.8331978982686996,
+      "num_tokens": 774009.0,
+      "step": 400
+    },
+    {
+      "epoch": 1.0,
+      "eval_entropy": 0.5918735759536715,
+      "eval_loss": 0.6536311507225037,
+      "eval_mean_token_accuracy": 0.8276140002438622,
+      "eval_num_tokens": 847606.0,
+      "eval_runtime": 82.2575,
+      "eval_samples_per_second": 13.251,
+      "eval_steps_per_second": 1.666,
+      "step": 438
+    },
+    {
+      "entropy": 0.654371697890577,
+      "epoch": 1.0274442538593482,
+      "grad_norm": 0.923791229724884,
+      "learning_rate": 0.00039553187999846057,
+      "loss": 0.5879265594482422,
+      "mean_token_accuracy": 0.8356293612325252,
+      "num_tokens": 870214.0,
+      "step": 450
+    },
+    {
+      "entropy": 0.6167843751609325,
+      "epoch": 1.1417953116066324,
+      "grad_norm": 0.7943944334983826,
+      "learning_rate": 0.00039530582686207966,
+      "loss": 0.552466049194336,
+      "mean_token_accuracy": 0.8413123995065689,
+      "num_tokens": 971584.0,
+      "step": 500
+    },
+    {
+      "entropy": 0.6443017952144146,
+      "epoch": 1.2561463693539165,
+      "grad_norm": 1.018168568611145,
+      "learning_rate": 0.00039476615968757247,
+      "loss": 0.5773007583618164,
+      "mean_token_accuracy": 0.8377226310968399,
+      "num_tokens": 1063906.0,
+      "step": 550
+    },
+    {
+      "entropy": 0.6407185651361942,
+      "epoch": 1.3704974271012007,
+      "grad_norm": 0.9198986291885376,
+      "learning_rate": 0.00039391373526624385,
+      "loss": 0.572706184387207,
+      "mean_token_accuracy": 0.8367982304096222,
+      "num_tokens": 1159282.0,
+      "step": 600
+    },
+    {
+      "entropy": 0.6494526116549969,
+      "epoch": 1.4848484848484849,
+      "grad_norm": 1.0277528762817383,
+      "learning_rate": 0.00039274990693190134,
+      "loss": 0.578397102355957,
+      "mean_token_accuracy": 0.836915987432003,
+      "num_tokens": 1254998.0,
+      "step": 650
+    },
+    {
+      "entropy": 0.6396139825880528,
+      "epoch": 1.599199542595769,
+      "grad_norm": 0.8996608257293701,
+      "learning_rate": 0.0003912765224122631,
+      "loss": 0.5766008758544922,
+      "mean_token_accuracy": 0.8396538972854615,
+      "num_tokens": 1351877.0,
+      "step": 700
+    },
+    {
+      "entropy": 0.6245273813605309,
+      "epoch": 1.7135506003430532,
+      "grad_norm": 1.1999542713165283,
+      "learning_rate": 0.00038949592089545203,
+      "loss": 0.5718669509887695,
+      "mean_token_accuracy": 0.8406938418745995,
+      "num_tokens": 1448801.0,
+      "step": 750
+    },
+    {
+      "entropy": 0.6350119508802891,
+      "epoch": 1.8279016580903373,
+      "grad_norm": 0.9171736836433411,
+      "learning_rate": 0.00038741092931623213,
+      "loss": 0.5792041778564453,
+      "mean_token_accuracy": 0.8370065066218376,
+      "num_tokens": 1547040.0,
+      "step": 800
+    },
+    {
+      "entropy": 0.6106068694591522,
+      "epoch": 1.9422527158376215,
+      "grad_norm": 0.6869989037513733,
+      "learning_rate": 0.0003850248578678851,
+      "loss": 0.5748995590209961,
+      "mean_token_accuracy": 0.8383069199323654,
+      "num_tokens": 1646444.0,
+      "step": 850
+    },
+    {
+      "epoch": 2.0,
+      "eval_entropy": 0.5696834810893902,
+      "eval_loss": 0.6577129364013672,
+      "eval_mean_token_accuracy": 0.8307611911836332,
+      "eval_num_tokens": 1695212.0,
+      "eval_runtime": 81.951,
+      "eval_samples_per_second": 13.301,
+      "eval_steps_per_second": 1.672,
+      "step": 876
+    },
+    {
+      "entropy": 0.572038494088323,
+      "epoch": 2.0548885077186965,
+      "grad_norm": 0.7493197321891785,
+      "learning_rate": 0.00038234149474685104,
+      "loss": 0.5193290710449219,
+      "mean_token_accuracy": 0.8497188209277119,
+      "num_tokens": 1738782.0,
+      "step": 900
+    },
+    {
+      "entropy": 0.5478819286823273,
+      "epoch": 2.1692395654659804,
+      "grad_norm": 1.2549943923950195,
+      "learning_rate": 0.0003793651001384773,
+      "loss": 0.4793633270263672,
+      "mean_token_accuracy": 0.8575673401355743,
+      "num_tokens": 1837910.0,
+      "step": 950
+    },
+    {
+      "entropy": 0.5431128491461277,
+      "epoch": 2.283590623213265,
+      "grad_norm": 0.8713223338127136,
+      "learning_rate": 0.00037610039945342474,
+      "loss": 0.4849379348754883,
+      "mean_token_accuracy": 0.8562448808550834,
+      "num_tokens": 1937110.0,
+      "step": 1000
+    },
+    {
+      "entropy": 0.5498340383172036,
+      "epoch": 2.3979416809605487,
+      "grad_norm": 0.6592629551887512,
+      "learning_rate": 0.00037255257582546804,
+      "loss": 0.493680419921875,
+      "mean_token_accuracy": 0.854153895676136,
+      "num_tokens": 2036890.0,
+      "step": 1050
+    },
+    {
+      "entropy": 0.5554873237013817,
+      "epoch": 2.512292738707833,
+      "grad_norm": 0.7878748774528503,
+      "learning_rate": 0.0003687272618826019,
+      "loss": 0.4951488494873047,
+      "mean_token_accuracy": 0.8540353265404701,
+      "num_tokens": 2137239.0,
+      "step": 1100
+    },
+    {
+      "entropy": 0.552360344529152,
+      "epoch": 2.6266437964551175,
+      "grad_norm": 0.7380415797233582,
+      "learning_rate": 0.00036463053080451683,
+      "loss": 0.49989810943603513,
+      "mean_token_accuracy": 0.8533406674861908,
+      "num_tokens": 2231081.0,
+      "step": 1150
+    },
+    {
+      "entropy": 0.5345852719247342,
+      "epoch": 2.7409948542024014,
+      "grad_norm": 0.7600613832473755,
+      "learning_rate": 0.0003602688866806424,
+      "loss": 0.5011516189575196,
+      "mean_token_accuracy": 0.8527707865834236,
+      "num_tokens": 2325503.0,
+      "step": 1200
+    },
+    {
+      "entropy": 0.532319468408823,
+      "epoch": 2.8553459119496853,
+      "grad_norm": 0.8197540640830994,
+      "learning_rate": 0.000355649254184066,
+      "loss": 0.5023023605346679,
+      "mean_token_accuracy": 0.8530397671461105,
+      "num_tokens": 2420627.0,
+      "step": 1250
+    },
+    {
+      "entropy": 0.5198138700425625,
+      "epoch": 2.9696969696969697,
+      "grad_norm": 0.8660516142845154,
+      "learning_rate": 0.00035077896757771987,
+      "loss": 0.4941487121582031,
+      "mean_token_accuracy": 0.855004439651966,
+      "num_tokens": 2518051.0,
+      "step": 1300
+    },
+    {
+      "epoch": 3.0,
+      "eval_entropy": 0.5642598084289662,
+      "eval_loss": 0.6125267744064331,
+      "eval_mean_token_accuracy": 0.8375998204641969,
+      "eval_num_tokens": 2542818.0,
+      "eval_runtime": 81.5524,
+      "eval_samples_per_second": 13.366,
+      "eval_steps_per_second": 1.68,
+      "step": 1314
+    },
+    {
+      "entropy": 0.45835615868495805,
+      "epoch": 3.0823327615780447,
+      "grad_norm": 0.734519362449646,
+      "learning_rate": 0.00034566575907029245,
+      "loss": 0.4334117889404297,
+      "mean_token_accuracy": 0.8698135275526095,
+      "num_tokens": 2613861.0,
+      "step": 1350
+    },
+    {
+      "entropy": 0.44648877054452896,
+      "epoch": 3.1966838193253286,
+      "grad_norm": 0.687981128692627,
+      "learning_rate": 0.00034031774654034867,
+      "loss": 0.40152584075927733,
+      "mean_token_accuracy": 0.8759277358651161,
+      "num_tokens": 2709451.0,
+      "step": 1400
+    },
+    {
+      "entropy": 0.4593352344632149,
+      "epoch": 3.311034877072613,
+      "grad_norm": 0.6310222148895264,
+      "learning_rate": 0.0003347434206481497,
+      "loss": 0.41885005950927734,
+      "mean_token_accuracy": 0.8714999267458916,
+      "num_tokens": 2801336.0,
+      "step": 1450
+    },
+    {
+      "entropy": 0.46372457168996334,
+      "epoch": 3.425385934819897,
+      "grad_norm": 0.6795068979263306,
+      "learning_rate": 0.0003289516313556339,
+      "loss": 0.4194014358520508,
+      "mean_token_accuracy": 0.8709167090058326,
+      "num_tokens": 2898945.0,
+      "step": 1500
+    },
+    {
+      "entropy": 0.4548304344713688,
+      "epoch": 3.5397369925671813,
+      "grad_norm": 0.9671820402145386,
+      "learning_rate": 0.00032295157387595997,
+      "loss": 0.4178035736083984,
+      "mean_token_accuracy": 0.8710319435596466,
+      "num_tokens": 2993378.0,
+      "step": 1550
+    },
+    {
+      "entropy": 0.4532127542793751,
+      "epoch": 3.6540880503144653,
+      "grad_norm": 0.7656959295272827,
+      "learning_rate": 0.0003167527740749187,
+      "loss": 0.42567527770996094,
+      "mean_token_accuracy": 0.869033499956131,
+      "num_tokens": 3091323.0,
+      "step": 1600
+    },
+    {
+      "entropy": 0.4460975781828165,
+      "epoch": 3.7684391080617496,
+      "grad_norm": 0.7811372876167297,
+      "learning_rate": 0.0003103650733473916,
+      "loss": 0.4148756790161133,
+      "mean_token_accuracy": 0.8709280017018318,
+      "num_tokens": 3193194.0,
+      "step": 1650
+    },
+    {
+      "entropy": 0.47125090911984446,
+      "epoch": 3.8827901658090336,
+      "grad_norm": 0.8664485216140747,
+      "learning_rate": 0.00030379861299286587,
+      "loss": 0.4209564590454102,
+      "mean_token_accuracy": 0.8695067319273949,
+      "num_tokens": 3292840.0,
+      "step": 1700
+    },
+    {
+      "entropy": 0.4641740974783897,
+      "epoch": 3.997141223556318,
+      "grad_norm": 0.85616534948349,
+      "learning_rate": 0.00029706381811481263,
+      "loss": 0.4319478988647461,
+      "mean_token_accuracy": 0.8689105224609375,
+      "num_tokens": 3387511.0,
+      "step": 1750
+    },
+    {
+      "epoch": 4.0,
+      "eval_entropy": 0.4623802898138979,
+      "eval_loss": 0.6707634925842285,
+      "eval_mean_token_accuracy": 0.8327476525828786,
+      "eval_num_tokens": 3390424.0,
+      "eval_runtime": 81.9106,
+      "eval_samples_per_second": 13.307,
+      "eval_steps_per_second": 1.673,
+      "step": 1752
+    },
+    {
+      "entropy": 0.33750920285125674,
+      "epoch": 4.109777015437393,
+      "grad_norm": 0.6137083172798157,
+      "learning_rate": 0.00029017138106948915,
+      "loss": 0.31130544662475584,
+      "mean_token_accuracy": 0.8997744660692166,
+      "num_tokens": 3483795.0,
+      "step": 1800
+    },
+    {
+      "entropy": 0.3504758469760418,
+      "epoch": 4.224128073184677,
+      "grad_norm": 0.6796069741249084,
+      "learning_rate": 0.00028313224449044337,
+      "loss": 0.31956762313842774,
+      "mean_token_accuracy": 0.895492787361145,
+      "num_tokens": 3576928.0,
+      "step": 1850
+    },
+    {
+      "entropy": 0.34297160781919955,
+      "epoch": 4.338479130931961,
+      "grad_norm": 0.7507141828536987,
+      "learning_rate": 0.00027595758391566995,
+      "loss": 0.31722068786621094,
+      "mean_token_accuracy": 0.8957494455575943,
+      "num_tokens": 3674125.0,
+      "step": 1900
+    },
+    {
+      "entropy": 0.3572557730227709,
+      "epoch": 4.452830188679245,
+      "grad_norm": 0.7509586215019226,
+      "learning_rate": 0.00026865879004500166,
+      "loss": 0.32728019714355466,
+      "mean_token_accuracy": 0.894504280090332,
+      "num_tokens": 3769853.0,
+      "step": 1950
+    },
+    {
+      "entropy": 0.35588217943906786,
+      "epoch": 4.56718124642653,
+      "grad_norm": 0.6168456077575684,
+      "learning_rate": 0.0002612474506559027,
+      "loss": 0.3304343795776367,
+      "mean_token_accuracy": 0.8935999310016632,
+      "num_tokens": 3867574.0,
+      "step": 2000
+    },
+    {
+      "entropy": 0.3554471937566996,
+      "epoch": 4.681532304173814,
+      "grad_norm": 0.6922969222068787,
+      "learning_rate": 0.00025373533220637623,
+      "loss": 0.3315126037597656,
+      "mean_token_accuracy": 0.8925720381736756,
+      "num_tokens": 3969484.0,
+      "step": 2050
+    },
+    {
+      "entropy": 0.35552985966205597,
+      "epoch": 4.795883361921097,
+      "grad_norm": 0.6013802289962769,
+      "learning_rate": 0.00024613436115419326,
+      "loss": 0.3315456771850586,
+      "mean_token_accuracy": 0.8929180511832238,
+      "num_tokens": 4063480.0,
+      "step": 2100
+    },
+    {
+      "entropy": 0.3478537338972092,
+      "epoch": 4.910234419668382,
+      "grad_norm": 0.8090653419494629,
+      "learning_rate": 0.00023845660502210138,
+      "loss": 0.32579368591308594,
+      "mean_token_accuracy": 0.893961855173111,
+      "num_tokens": 4162925.0,
+      "step": 2150
+    },
+    {
+      "epoch": 5.0,
+      "eval_entropy": 0.38475005339532004,
+      "eval_loss": 0.6992377042770386,
+      "eval_mean_token_accuracy": 0.8384414993933518,
+      "eval_num_tokens": 4238030.0,
+      "eval_runtime": 81.1991,
+      "eval_samples_per_second": 13.424,
+      "eval_steps_per_second": 1.687,
+      "step": 2190
+    },
+    {
+      "entropy": 0.3381291570397198,
+      "epoch": 5.022870211549457,
+      "grad_norm": 0.7052969932556152,
+      "learning_rate": 0.00023071425323907478,
+      "loss": 0.3124395370483398,
+      "mean_token_accuracy": 0.8997272504162668,
+      "num_tokens": 4257610.0,
+      "step": 2200
+    },
+    {
+      "entropy": 0.2494013351202011,
+      "epoch": 5.137221269296741,
+      "grad_norm": 0.8653550148010254,
+      "learning_rate": 0.0002229195977880219,
+      "loss": 0.21398515701293946,
+      "mean_token_accuracy": 0.9277590665221215,
+      "num_tokens": 4353703.0,
+      "step": 2250
+    },
+    {
+      "entropy": 0.27507017262279987,
+      "epoch": 5.251572327044025,
+      "grad_norm": 0.7887904644012451,
+      "learning_rate": 0.00021508501369067626,
+      "loss": 0.22058496475219727,
+      "mean_token_accuracy": 0.9253441894054413,
+      "num_tokens": 4449829.0,
+      "step": 2300
+    },
+    {
+      "entropy": 0.2582047886401415,
+      "epoch": 5.3659233847913095,
+      "grad_norm": 0.8728304505348206,
+      "learning_rate": 0.0002072229393606514,
+      "loss": 0.22437936782836915,
+      "mean_token_accuracy": 0.9235965391993523,
+      "num_tokens": 4545644.0,
+      "step": 2350
+    },
+    {
+      "entropy": 0.26232284381985665,
+      "epoch": 5.480274442538594,
+      "grad_norm": 0.6688716411590576,
+      "learning_rate": 0.00019934585685585428,
+      "loss": 0.23260833740234374,
+      "mean_token_accuracy": 0.9217587423324585,
+      "num_tokens": 4637941.0,
+      "step": 2400
+    },
+    {
+      "entropy": 0.2619731902331114,
+      "epoch": 5.594625500285877,
+      "grad_norm": 0.597118616104126,
+      "learning_rate": 0.00019146627206160645,
+      "loss": 0.22075172424316405,
+      "mean_token_accuracy": 0.9252614346146584,
+      "num_tokens": 4738175.0,
+      "step": 2450
+    },
+    {
+      "entropy": 0.26591138120740654,
+      "epoch": 5.708976558033162,
+      "grad_norm": 0.7602102160453796,
+      "learning_rate": 0.00018359669483593703,
+      "loss": 0.22200309753417968,
+      "mean_token_accuracy": 0.9250017151236534,
+      "num_tokens": 4837766.0,
+      "step": 2500
+    },
+    {
+      "entropy": 0.2587306099012494,
+      "epoch": 5.823327615780446,
+      "grad_norm": 0.6875442862510681,
+      "learning_rate": 0.00017574961914856833,
+      "loss": 0.22627996444702148,
+      "mean_token_accuracy": 0.9234379661083222,
+      "num_tokens": 4935670.0,
+      "step": 2550
+    },
+    {
+      "entropy": 0.2689222282171249,
+      "epoch": 5.9376786735277305,
+      "grad_norm": 0.654289186000824,
+      "learning_rate": 0.00016793750324512556,
+      "loss": 0.23186542510986327,
+      "mean_token_accuracy": 0.9214966455101967,
+      "num_tokens": 5030799.0,
+      "step": 2600
+    },
+    {
+      "epoch": 6.0,
+      "eval_entropy": 0.34485617432281046,
+      "eval_loss": 0.7253938317298889,
+      "eval_mean_token_accuracy": 0.8377144193127207,
+      "eval_num_tokens": 5085636.0,
+      "eval_runtime": 81.5281,
+      "eval_samples_per_second": 13.37,
+      "eval_steps_per_second": 1.68,
+      "step": 2628
+    },
+    {
+      "entropy": 0.22255296551334072,
+      "epoch": 6.050314465408805,
+      "grad_norm": 0.5370154976844788,
+      "learning_rate": 0.00016017274986806444,
+      "loss": 0.18560497283935548,
+      "mean_token_accuracy": 0.9371391822238864,
+      "num_tokens": 5126968.0,
+      "step": 2650
+    },
+    {
+      "entropy": 0.1745598777383566,
+      "epoch": 6.164665523156089,
+      "grad_norm": 0.6606088280677795,
+      "learning_rate": 0.00015246768656571565,
+      "loss": 0.14104312896728516,
+      "mean_token_accuracy": 0.9522112870216369,
+      "num_tokens": 5222926.0,
+      "step": 2700
+    },
+    {
+      "entropy": 0.16993012264370919,
+      "epoch": 6.279016580903374,
+      "grad_norm": 0.6963379383087158,
+      "learning_rate": 0.00014483454612071133,
+      "loss": 0.14232961654663087,
+      "mean_token_accuracy": 0.9532496619224549,
+      "num_tokens": 5317743.0,
+      "step": 2750
+    },
+    {
+      "entropy": 0.16726660918444394,
+      "epoch": 6.393367638650657,
+      "grad_norm": 0.7356508374214172,
+      "learning_rate": 0.00013728544712886273,
+      "loss": 0.13951081275939942,
+      "mean_token_accuracy": 0.9533490630984306,
+      "num_tokens": 5415584.0,
+      "step": 2800
+    },
+    {
+      "entropy": 0.17247022703289985,
+      "epoch": 6.507718696397942,
+      "grad_norm": 0.6522324681282043,
+      "learning_rate": 0.00012983237475932584,
+      "loss": 0.13756629943847656,
+      "mean_token_accuracy": 0.9529287284612655,
+      "num_tokens": 5516998.0,
+      "step": 2850
+    },
+    {
+      "entropy": 0.1730361183360219,
+      "epoch": 6.622069754145226,
+      "grad_norm": 0.747295618057251,
+      "learning_rate": 0.00012248716172659724,
+      "loss": 0.1429327392578125,
+      "mean_token_accuracy": 0.9519825592637062,
+      "num_tokens": 5614424.0,
+      "step": 2900
+    },
+    {
+      "entropy": 0.17447776831686496,
+      "epoch": 6.7364208118925095,
+      "grad_norm": 0.624346137046814,
+      "learning_rate": 0.00011526146950455331,
+      "loss": 0.14175209999084473,
+      "mean_token_accuracy": 0.9515316143631936,
+      "num_tokens": 5712047.0,
+      "step": 2950
+    },
+    {
+      "entropy": 0.181520360969007,
+      "epoch": 6.850771869639794,
+      "grad_norm": 0.747399091720581,
+      "learning_rate": 0.00010816676981235426,
+      "loss": 0.14949700355529785,
+      "mean_token_accuracy": 0.9500517123937606,
+      "num_tokens": 5805524.0,
+      "step": 3000
+    },
+    {
+      "entropy": 0.17918075375258924,
+      "epoch": 6.965122927387078,
+      "grad_norm": 0.55885249376297,
+      "learning_rate": 0.00010121432640160985,
+      "loss": 0.14284159660339354,
+      "mean_token_accuracy": 0.9515545222163201,
+      "num_tokens": 5902110.0,
+      "step": 3050
+    },
+    {
+      "epoch": 7.0,
+      "eval_entropy": 0.2829361812476694,
+      "eval_loss": 0.8297234177589417,
+      "eval_mean_token_accuracy": 0.8362262305552072,
+      "eval_num_tokens": 5933242.0,
+      "eval_runtime": 81.6661,
+      "eval_samples_per_second": 13.347,
+      "eval_steps_per_second": 1.678,
+      "step": 3066
+    }
+  ],
+  "logging_steps": 50,
+  "max_steps": 4380,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 10,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 2.4648107843557786e+17,
+  "train_batch_size": 4,
+  "trial_name": null,
+  "trial_params": null
+}

systematicity_original_Swedish/Qwen3.5-4B-Base_systematicity_splits_original_features_train_systematicity_splits_original_features_test1/checkpoint-3504/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

systematicity_original_Swedish/Qwen3.5-4B-Base_systematicity_splits_original_features_train_systematicity_splits_original_features_test1/checkpoint-3504/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": 0.026477173062682283,
+  "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": [
+    "q_proj",
+    "k_proj",
+    "down_proj",
+    "v_proj",
+    "gate_proj",
+    "o_proj",
+    "up_proj"
+  ],
+  "target_parameters": null,
+  "task_type": "CAUSAL_LM",
+  "trainable_token_indices": null,
+  "use_dora": false,
+  "use_qalora": false,
+  "use_rslora": false
+}

systematicity_original_Swedish/Qwen3.5-4B-Base_systematicity_splits_original_features_train_systematicity_splits_original_features_test1/checkpoint-3504/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 %}

systematicity_original_Swedish/Qwen3.5-4B-Base_systematicity_splits_original_features_train_systematicity_splits_original_features_test1/checkpoint-3504/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|>"
+}

systematicity_original_Swedish/Qwen3.5-4B-Base_systematicity_splits_original_features_train_systematicity_splits_original_features_test1/checkpoint-3504/trainer_state.json

@@ -0,0 +1,822 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 8.0,
+  "eval_steps": 500,
+  "global_step": 3504,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "entropy": 1.4168990224599838,
+      "epoch": 0.11435105774728416,
+      "grad_norm": 1.3415299654006958,
+      "learning_rate": 4.424985043427545e-05,
+      "loss": 1.2160142517089845,
+      "mean_token_accuracy": 0.724469550549984,
+      "num_tokens": 96586.0,
+      "step": 50
+    },
+    {
+      "entropy": 0.7365697559714317,
+      "epoch": 0.22870211549456831,
+      "grad_norm": 0.9586130380630493,
+      "learning_rate": 8.940275904067897e-05,
+      "loss": 0.6445799255371094,
+      "mean_token_accuracy": 0.822787437736988,
+      "num_tokens": 190720.0,
+      "step": 100
+    },
+    {
+      "entropy": 0.691228820681572,
+      "epoch": 0.34305317324185247,
+      "grad_norm": 0.7875016331672668,
+      "learning_rate": 0.0001345556676470825,
+      "loss": 0.6011501312255859,
+      "mean_token_accuracy": 0.8325563898682594,
+      "num_tokens": 288515.0,
+      "step": 150
+    },
+    {
+      "entropy": 0.6706234656274319,
+      "epoch": 0.45740423098913663,
+      "grad_norm": 0.7867940068244934,
+      "learning_rate": 0.000179708576253486,
+      "loss": 0.5852980422973633,
+      "mean_token_accuracy": 0.83610415995121,
+      "num_tokens": 387874.0,
+      "step": 200
+    },
+    {
+      "entropy": 0.6558432649075985,
+      "epoch": 0.5717552887364208,
+      "grad_norm": 0.8049650192260742,
+      "learning_rate": 0.00022486148485988957,
+      "loss": 0.5714554977416992,
+      "mean_token_accuracy": 0.8380795314908027,
+      "num_tokens": 487865.0,
+      "step": 250
+    },
+    {
+      "entropy": 0.6612135827541351,
+      "epoch": 0.6861063464837049,
+      "grad_norm": 0.6776595711708069,
+      "learning_rate": 0.0002700143934662931,
+      "loss": 0.5815717315673828,
+      "mean_token_accuracy": 0.8377441626787185,
+      "num_tokens": 584269.0,
+      "step": 300
+    },
+    {
+      "entropy": 0.6668009972572326,
+      "epoch": 0.8004574042309891,
+      "grad_norm": 1.132367491722107,
+      "learning_rate": 0.0003151673020726966,
+      "loss": 0.5866946411132813,
+      "mean_token_accuracy": 0.8364784654974937,
+      "num_tokens": 676067.0,
+      "step": 350
+    },
+    {
+      "entropy": 0.6667388904094697,
+      "epoch": 0.9148084619782733,
+      "grad_norm": 0.9101859331130981,
+      "learning_rate": 0.0003603202106791001,
+      "loss": 0.5946685028076172,
+      "mean_token_accuracy": 0.8331978982686996,
+      "num_tokens": 774009.0,
+      "step": 400
+    },
+    {
+      "epoch": 1.0,
+      "eval_entropy": 0.5918735759536715,
+      "eval_loss": 0.6536311507225037,
+      "eval_mean_token_accuracy": 0.8276140002438622,
+      "eval_num_tokens": 847606.0,
+      "eval_runtime": 82.2575,
+      "eval_samples_per_second": 13.251,
+      "eval_steps_per_second": 1.666,
+      "step": 438
+    },
+    {
+      "entropy": 0.654371697890577,
+      "epoch": 1.0274442538593482,
+      "grad_norm": 0.923791229724884,
+      "learning_rate": 0.00039553187999846057,
+      "loss": 0.5879265594482422,
+      "mean_token_accuracy": 0.8356293612325252,
+      "num_tokens": 870214.0,
+      "step": 450
+    },
+    {
+      "entropy": 0.6167843751609325,
+      "epoch": 1.1417953116066324,
+      "grad_norm": 0.7943944334983826,
+      "learning_rate": 0.00039530582686207966,
+      "loss": 0.552466049194336,
+      "mean_token_accuracy": 0.8413123995065689,
+      "num_tokens": 971584.0,
+      "step": 500
+    },
+    {
+      "entropy": 0.6443017952144146,
+      "epoch": 1.2561463693539165,
+      "grad_norm": 1.018168568611145,
+      "learning_rate": 0.00039476615968757247,
+      "loss": 0.5773007583618164,
+      "mean_token_accuracy": 0.8377226310968399,
+      "num_tokens": 1063906.0,
+      "step": 550
+    },
+    {
+      "entropy": 0.6407185651361942,
+      "epoch": 1.3704974271012007,
+      "grad_norm": 0.9198986291885376,
+      "learning_rate": 0.00039391373526624385,
+      "loss": 0.572706184387207,
+      "mean_token_accuracy": 0.8367982304096222,
+      "num_tokens": 1159282.0,
+      "step": 600
+    },
+    {
+      "entropy": 0.6494526116549969,
+      "epoch": 1.4848484848484849,
+      "grad_norm": 1.0277528762817383,
+      "learning_rate": 0.00039274990693190134,
+      "loss": 0.578397102355957,
+      "mean_token_accuracy": 0.836915987432003,
+      "num_tokens": 1254998.0,
+      "step": 650
+    },
+    {
+      "entropy": 0.6396139825880528,
+      "epoch": 1.599199542595769,
+      "grad_norm": 0.8996608257293701,
+      "learning_rate": 0.0003912765224122631,
+      "loss": 0.5766008758544922,
+      "mean_token_accuracy": 0.8396538972854615,
+      "num_tokens": 1351877.0,
+      "step": 700
+    },
+    {
+      "entropy": 0.6245273813605309,
+      "epoch": 1.7135506003430532,
+      "grad_norm": 1.1999542713165283,
+      "learning_rate": 0.00038949592089545203,
+      "loss": 0.5718669509887695,
+      "mean_token_accuracy": 0.8406938418745995,
+      "num_tokens": 1448801.0,
+      "step": 750
+    },
+    {
+      "entropy": 0.6350119508802891,
+      "epoch": 1.8279016580903373,
+      "grad_norm": 0.9171736836433411,
+      "learning_rate": 0.00038741092931623213,
+      "loss": 0.5792041778564453,
+      "mean_token_accuracy": 0.8370065066218376,
+      "num_tokens": 1547040.0,
+      "step": 800
+    },
+    {
+      "entropy": 0.6106068694591522,
+      "epoch": 1.9422527158376215,
+      "grad_norm": 0.6869989037513733,
+      "learning_rate": 0.0003850248578678851,
+      "loss": 0.5748995590209961,
+      "mean_token_accuracy": 0.8383069199323654,
+      "num_tokens": 1646444.0,
+      "step": 850
+    },
+    {
+      "epoch": 2.0,
+      "eval_entropy": 0.5696834810893902,
+      "eval_loss": 0.6577129364013672,
+      "eval_mean_token_accuracy": 0.8307611911836332,
+      "eval_num_tokens": 1695212.0,
+      "eval_runtime": 81.951,
+      "eval_samples_per_second": 13.301,
+      "eval_steps_per_second": 1.672,
+      "step": 876
+    },
+    {
+      "entropy": 0.572038494088323,
+      "epoch": 2.0548885077186965,
+      "grad_norm": 0.7493197321891785,
+      "learning_rate": 0.00038234149474685104,
+      "loss": 0.5193290710449219,
+      "mean_token_accuracy": 0.8497188209277119,
+      "num_tokens": 1738782.0,
+      "step": 900
+    },
+    {
+      "entropy": 0.5478819286823273,
+      "epoch": 2.1692395654659804,
+      "grad_norm": 1.2549943923950195,
+      "learning_rate": 0.0003793651001384773,
+      "loss": 0.4793633270263672,
+      "mean_token_accuracy": 0.8575673401355743,
+      "num_tokens": 1837910.0,
+      "step": 950
+    },
+    {
+      "entropy": 0.5431128491461277,
+      "epoch": 2.283590623213265,
+      "grad_norm": 0.8713223338127136,
+      "learning_rate": 0.00037610039945342474,
+      "loss": 0.4849379348754883,
+      "mean_token_accuracy": 0.8562448808550834,
+      "num_tokens": 1937110.0,
+      "step": 1000
+    },
+    {
+      "entropy": 0.5498340383172036,
+      "epoch": 2.3979416809605487,
+      "grad_norm": 0.6592629551887512,
+      "learning_rate": 0.00037255257582546804,
+      "loss": 0.493680419921875,
+      "mean_token_accuracy": 0.854153895676136,
+      "num_tokens": 2036890.0,
+      "step": 1050
+    },
+    {
+      "entropy": 0.5554873237013817,
+      "epoch": 2.512292738707833,
+      "grad_norm": 0.7878748774528503,
+      "learning_rate": 0.0003687272618826019,
+      "loss": 0.4951488494873047,
+      "mean_token_accuracy": 0.8540353265404701,
+      "num_tokens": 2137239.0,
+      "step": 1100
+    },
+    {
+      "entropy": 0.552360344529152,
+      "epoch": 2.6266437964551175,
+      "grad_norm": 0.7380415797233582,
+      "learning_rate": 0.00036463053080451683,
+      "loss": 0.49989810943603513,
+      "mean_token_accuracy": 0.8533406674861908,
+      "num_tokens": 2231081.0,
+      "step": 1150
+    },
+    {
+      "entropy": 0.5345852719247342,
+      "epoch": 2.7409948542024014,
+      "grad_norm": 0.7600613832473755,
+      "learning_rate": 0.0003602688866806424,
+      "loss": 0.5011516189575196,
+      "mean_token_accuracy": 0.8527707865834236,
+      "num_tokens": 2325503.0,
+      "step": 1200
+    },
+    {
+      "entropy": 0.532319468408823,
+      "epoch": 2.8553459119496853,
+      "grad_norm": 0.8197540640830994,
+      "learning_rate": 0.000355649254184066,
+      "loss": 0.5023023605346679,
+      "mean_token_accuracy": 0.8530397671461105,
+      "num_tokens": 2420627.0,
+      "step": 1250
+    },
+    {
+      "entropy": 0.5198138700425625,
+      "epoch": 2.9696969696969697,
+      "grad_norm": 0.8660516142845154,
+      "learning_rate": 0.00035077896757771987,
+      "loss": 0.4941487121582031,
+      "mean_token_accuracy": 0.855004439651966,
+      "num_tokens": 2518051.0,
+      "step": 1300
+    },
+    {
+      "epoch": 3.0,
+      "eval_entropy": 0.5642598084289662,
+      "eval_loss": 0.6125267744064331,
+      "eval_mean_token_accuracy": 0.8375998204641969,
+      "eval_num_tokens": 2542818.0,
+      "eval_runtime": 81.5524,
+      "eval_samples_per_second": 13.366,
+      "eval_steps_per_second": 1.68,
+      "step": 1314
+    },
+    {
+      "entropy": 0.45835615868495805,
+      "epoch": 3.0823327615780447,
+      "grad_norm": 0.734519362449646,
+      "learning_rate": 0.00034566575907029245,
+      "loss": 0.4334117889404297,
+      "mean_token_accuracy": 0.8698135275526095,
+      "num_tokens": 2613861.0,
+      "step": 1350
+    },
+    {
+      "entropy": 0.44648877054452896,
+      "epoch": 3.1966838193253286,
+      "grad_norm": 0.687981128692627,
+      "learning_rate": 0.00034031774654034867,
+      "loss": 0.40152584075927733,
+      "mean_token_accuracy": 0.8759277358651161,
+      "num_tokens": 2709451.0,
+      "step": 1400
+    },
+    {
+      "entropy": 0.4593352344632149,
+      "epoch": 3.311034877072613,
+      "grad_norm": 0.6310222148895264,
+      "learning_rate": 0.0003347434206481497,
+      "loss": 0.41885005950927734,
+      "mean_token_accuracy": 0.8714999267458916,
+      "num_tokens": 2801336.0,
+      "step": 1450
+    },
+    {
+      "entropy": 0.46372457168996334,
+      "epoch": 3.425385934819897,
+      "grad_norm": 0.6795068979263306,
+      "learning_rate": 0.0003289516313556339,
+      "loss": 0.4194014358520508,
+      "mean_token_accuracy": 0.8709167090058326,
+      "num_tokens": 2898945.0,
+      "step": 1500
+    },
+    {
+      "entropy": 0.4548304344713688,
+      "epoch": 3.5397369925671813,
+      "grad_norm": 0.9671820402145386,
+      "learning_rate": 0.00032295157387595997,
+      "loss": 0.4178035736083984,
+      "mean_token_accuracy": 0.8710319435596466,
+      "num_tokens": 2993378.0,
+      "step": 1550
+    },
+    {
+      "entropy": 0.4532127542793751,
+      "epoch": 3.6540880503144653,
+      "grad_norm": 0.7656959295272827,
+      "learning_rate": 0.0003167527740749187,
+      "loss": 0.42567527770996094,
+      "mean_token_accuracy": 0.869033499956131,
+      "num_tokens": 3091323.0,
+      "step": 1600
+    },
+    {
+      "entropy": 0.4460975781828165,
+      "epoch": 3.7684391080617496,
+      "grad_norm": 0.7811372876167297,
+      "learning_rate": 0.0003103650733473916,
+      "loss": 0.4148756790161133,
+      "mean_token_accuracy": 0.8709280017018318,
+      "num_tokens": 3193194.0,
+      "step": 1650
+    },
+    {
+      "entropy": 0.47125090911984446,
+      "epoch": 3.8827901658090336,
+      "grad_norm": 0.8664485216140747,
+      "learning_rate": 0.00030379861299286587,
+      "loss": 0.4209564590454102,
+      "mean_token_accuracy": 0.8695067319273949,
+      "num_tokens": 3292840.0,
+      "step": 1700
+    },
+    {
+      "entropy": 0.4641740974783897,
+      "epoch": 3.997141223556318,
+      "grad_norm": 0.85616534948349,
+      "learning_rate": 0.00029706381811481263,
+      "loss": 0.4319478988647461,
+      "mean_token_accuracy": 0.8689105224609375,
+      "num_tokens": 3387511.0,
+      "step": 1750
+    },
+    {
+      "epoch": 4.0,
+      "eval_entropy": 0.4623802898138979,
+      "eval_loss": 0.6707634925842285,
+      "eval_mean_token_accuracy": 0.8327476525828786,
+      "eval_num_tokens": 3390424.0,
+      "eval_runtime": 81.9106,
+      "eval_samples_per_second": 13.307,
+      "eval_steps_per_second": 1.673,
+      "step": 1752
+    },
+    {
+      "entropy": 0.33750920285125674,
+      "epoch": 4.109777015437393,
+      "grad_norm": 0.6137083172798157,
+      "learning_rate": 0.00029017138106948915,
+      "loss": 0.31130544662475584,
+      "mean_token_accuracy": 0.8997744660692166,
+      "num_tokens": 3483795.0,
+      "step": 1800
+    },
+    {
+      "entropy": 0.3504758469760418,
+      "epoch": 4.224128073184677,
+      "grad_norm": 0.6796069741249084,
+      "learning_rate": 0.00028313224449044337,
+      "loss": 0.31956762313842774,
+      "mean_token_accuracy": 0.895492787361145,
+      "num_tokens": 3576928.0,
+      "step": 1850
+    },
+    {
+      "entropy": 0.34297160781919955,
+      "epoch": 4.338479130931961,
+      "grad_norm": 0.7507141828536987,
+      "learning_rate": 0.00027595758391566995,
+      "loss": 0.31722068786621094,
+      "mean_token_accuracy": 0.8957494455575943,
+      "num_tokens": 3674125.0,
+      "step": 1900
+    },
+    {
+      "entropy": 0.3572557730227709,
+      "epoch": 4.452830188679245,
+      "grad_norm": 0.7509586215019226,
+      "learning_rate": 0.00026865879004500166,
+      "loss": 0.32728019714355466,
+      "mean_token_accuracy": 0.894504280090332,
+      "num_tokens": 3769853.0,
+      "step": 1950
+    },
+    {
+      "entropy": 0.35588217943906786,
+      "epoch": 4.56718124642653,
+      "grad_norm": 0.6168456077575684,
+      "learning_rate": 0.0002612474506559027,
+      "loss": 0.3304343795776367,
+      "mean_token_accuracy": 0.8935999310016632,
+      "num_tokens": 3867574.0,
+      "step": 2000
+    },
+    {
+      "entropy": 0.3554471937566996,
+      "epoch": 4.681532304173814,
+      "grad_norm": 0.6922969222068787,
+      "learning_rate": 0.00025373533220637623,
+      "loss": 0.3315126037597656,
+      "mean_token_accuracy": 0.8925720381736756,
+      "num_tokens": 3969484.0,
+      "step": 2050
+    },
+    {
+      "entropy": 0.35552985966205597,
+      "epoch": 4.795883361921097,
+      "grad_norm": 0.6013802289962769,
+      "learning_rate": 0.00024613436115419326,
+      "loss": 0.3315456771850586,
+      "mean_token_accuracy": 0.8929180511832238,
+      "num_tokens": 4063480.0,
+      "step": 2100
+    },
+    {
+      "entropy": 0.3478537338972092,
+      "epoch": 4.910234419668382,
+      "grad_norm": 0.8090653419494629,
+      "learning_rate": 0.00023845660502210138,
+      "loss": 0.32579368591308594,
+      "mean_token_accuracy": 0.893961855173111,
+      "num_tokens": 4162925.0,
+      "step": 2150
+    },
+    {
+      "epoch": 5.0,
+      "eval_entropy": 0.38475005339532004,
+      "eval_loss": 0.6992377042770386,
+      "eval_mean_token_accuracy": 0.8384414993933518,
+      "eval_num_tokens": 4238030.0,
+      "eval_runtime": 81.1991,
+      "eval_samples_per_second": 13.424,
+      "eval_steps_per_second": 1.687,
+      "step": 2190
+    },
+    {
+      "entropy": 0.3381291570397198,
+      "epoch": 5.022870211549457,
+      "grad_norm": 0.7052969932556152,
+      "learning_rate": 0.00023071425323907478,
+      "loss": 0.3124395370483398,
+      "mean_token_accuracy": 0.8997272504162668,
+      "num_tokens": 4257610.0,
+      "step": 2200
+    },
+    {
+      "entropy": 0.2494013351202011,
+      "epoch": 5.137221269296741,
+      "grad_norm": 0.8653550148010254,
+      "learning_rate": 0.0002229195977880219,
+      "loss": 0.21398515701293946,
+      "mean_token_accuracy": 0.9277590665221215,
+      "num_tokens": 4353703.0,
+      "step": 2250
+    },
+    {
+      "entropy": 0.27507017262279987,
+      "epoch": 5.251572327044025,
+      "grad_norm": 0.7887904644012451,
+      "learning_rate": 0.00021508501369067626,
+      "loss": 0.22058496475219727,
+      "mean_token_accuracy": 0.9253441894054413,
+      "num_tokens": 4449829.0,
+      "step": 2300
+    },
+    {
+      "entropy": 0.2582047886401415,
+      "epoch": 5.3659233847913095,
+      "grad_norm": 0.8728304505348206,
+      "learning_rate": 0.0002072229393606514,
+      "loss": 0.22437936782836915,
+      "mean_token_accuracy": 0.9235965391993523,
+      "num_tokens": 4545644.0,
+      "step": 2350
+    },
+    {
+      "entropy": 0.26232284381985665,
+      "epoch": 5.480274442538594,
+      "grad_norm": 0.6688716411590576,
+      "learning_rate": 0.00019934585685585428,
+      "loss": 0.23260833740234374,
+      "mean_token_accuracy": 0.9217587423324585,
+      "num_tokens": 4637941.0,
+      "step": 2400
+    },
+    {
+      "entropy": 0.2619731902331114,
+      "epoch": 5.594625500285877,
+      "grad_norm": 0.597118616104126,
+      "learning_rate": 0.00019146627206160645,
+      "loss": 0.22075172424316405,
+      "mean_token_accuracy": 0.9252614346146584,
+      "num_tokens": 4738175.0,
+      "step": 2450
+    },
+    {
+      "entropy": 0.26591138120740654,
+      "epoch": 5.708976558033162,
+      "grad_norm": 0.7602102160453796,
+      "learning_rate": 0.00018359669483593703,
+      "loss": 0.22200309753417968,
+      "mean_token_accuracy": 0.9250017151236534,
+      "num_tokens": 4837766.0,
+      "step": 2500
+    },
+    {
+      "entropy": 0.2587306099012494,
+      "epoch": 5.823327615780446,
+      "grad_norm": 0.6875442862510681,
+      "learning_rate": 0.00017574961914856833,
+      "loss": 0.22627996444702148,
+      "mean_token_accuracy": 0.9234379661083222,
+      "num_tokens": 4935670.0,
+      "step": 2550
+    },
+    {
+      "entropy": 0.2689222282171249,
+      "epoch": 5.9376786735277305,
+      "grad_norm": 0.654289186000824,
+      "learning_rate": 0.00016793750324512556,
+      "loss": 0.23186542510986327,
+      "mean_token_accuracy": 0.9214966455101967,
+      "num_tokens": 5030799.0,
+      "step": 2600
+    },
+    {
+      "epoch": 6.0,
+      "eval_entropy": 0.34485617432281046,
+      "eval_loss": 0.7253938317298889,
+      "eval_mean_token_accuracy": 0.8377144193127207,
+      "eval_num_tokens": 5085636.0,
+      "eval_runtime": 81.5281,
+      "eval_samples_per_second": 13.37,
+      "eval_steps_per_second": 1.68,
+      "step": 2628
+    },
+    {
+      "entropy": 0.22255296551334072,
+      "epoch": 6.050314465408805,
+      "grad_norm": 0.5370154976844788,
+      "learning_rate": 0.00016017274986806444,
+      "loss": 0.18560497283935548,
+      "mean_token_accuracy": 0.9371391822238864,
+      "num_tokens": 5126968.0,
+      "step": 2650
+    },
+    {
+      "entropy": 0.1745598777383566,
+      "epoch": 6.164665523156089,
+      "grad_norm": 0.6606088280677795,
+      "learning_rate": 0.00015246768656571565,
+      "loss": 0.14104312896728516,
+      "mean_token_accuracy": 0.9522112870216369,
+      "num_tokens": 5222926.0,
+      "step": 2700
+    },
+    {
+      "entropy": 0.16993012264370919,
+      "epoch": 6.279016580903374,
+      "grad_norm": 0.6963379383087158,
+      "learning_rate": 0.00014483454612071133,
+      "loss": 0.14232961654663087,
+      "mean_token_accuracy": 0.9532496619224549,
+      "num_tokens": 5317743.0,
+      "step": 2750
+    },
+    {
+      "entropy": 0.16726660918444394,
+      "epoch": 6.393367638650657,
+      "grad_norm": 0.7356508374214172,
+      "learning_rate": 0.00013728544712886273,
+      "loss": 0.13951081275939942,
+      "mean_token_accuracy": 0.9533490630984306,
+      "num_tokens": 5415584.0,
+      "step": 2800
+    },
+    {
+      "entropy": 0.17247022703289985,
+      "epoch": 6.507718696397942,
+      "grad_norm": 0.6522324681282043,
+      "learning_rate": 0.00012983237475932584,
+      "loss": 0.13756629943847656,
+      "mean_token_accuracy": 0.9529287284612655,
+      "num_tokens": 5516998.0,
+      "step": 2850
+    },
+    {
+      "entropy": 0.1730361183360219,
+      "epoch": 6.622069754145226,
+      "grad_norm": 0.747295618057251,
+      "learning_rate": 0.00012248716172659724,
+      "loss": 0.1429327392578125,
+      "mean_token_accuracy": 0.9519825592637062,
+      "num_tokens": 5614424.0,
+      "step": 2900
+    },
+    {
+      "entropy": 0.17447776831686496,
+      "epoch": 6.7364208118925095,
+      "grad_norm": 0.624346137046814,
+      "learning_rate": 0.00011526146950455331,
+      "loss": 0.14175209999084473,
+      "mean_token_accuracy": 0.9515316143631936,
+      "num_tokens": 5712047.0,
+      "step": 2950
+    },
+    {
+      "entropy": 0.181520360969007,
+      "epoch": 6.850771869639794,
+      "grad_norm": 0.747399091720581,
+      "learning_rate": 0.00010816676981235426,
+      "loss": 0.14949700355529785,
+      "mean_token_accuracy": 0.9500517123937606,
+      "num_tokens": 5805524.0,
+      "step": 3000
+    },
+    {
+      "entropy": 0.17918075375258924,
+      "epoch": 6.965122927387078,
+      "grad_norm": 0.55885249376297,
+      "learning_rate": 0.00010121432640160985,
+      "loss": 0.14284159660339354,
+      "mean_token_accuracy": 0.9515545222163201,
+      "num_tokens": 5902110.0,
+      "step": 3050
+    },
+    {
+      "epoch": 7.0,
+      "eval_entropy": 0.2829361812476694,
+      "eval_loss": 0.8297234177589417,
+      "eval_mean_token_accuracy": 0.8362262305552072,
+      "eval_num_tokens": 5933242.0,
+      "eval_runtime": 81.6661,
+      "eval_samples_per_second": 13.347,
+      "eval_steps_per_second": 1.678,
+      "step": 3066
+    },
+    {
+      "entropy": 0.13936093423118445,
+      "epoch": 7.077758719268153,
+      "grad_norm": 0.6394845843315125,
+      "learning_rate": 9.441517717371896e-05,
+      "loss": 0.11461298942565917,
+      "mean_token_accuracy": 0.9617519581378414,
+      "num_tokens": 5994998.0,
+      "step": 3100
+    },
+    {
+      "entropy": 0.12445554660633207,
+      "epoch": 7.192109777015437,
+      "grad_norm": 0.3524336814880371,
+      "learning_rate": 8.778011665577682e-05,
+      "loss": 0.09778708457946778,
+      "mean_token_accuracy": 0.9672333124279976,
+      "num_tokens": 6089591.0,
+      "step": 3150
+    },
+    {
+      "entropy": 0.11911977402865886,
+      "epoch": 7.306460834762722,
+      "grad_norm": 0.5336443185806274,
+      "learning_rate": 8.131967886286935e-05,
+      "loss": 0.0957869529724121,
+      "mean_token_accuracy": 0.9681107413768768,
+      "num_tokens": 6189134.0,
+      "step": 3200
+    },
+    {
+      "entropy": 0.12348919529467821,
+      "epoch": 7.420811892510006,
+      "grad_norm": 0.3701528310775757,
+      "learning_rate": 7.504412057396483e-05,
+      "loss": 0.09508245468139648,
+      "mean_token_accuracy": 0.9655187261104584,
+      "num_tokens": 6290034.0,
+      "step": 3250
+    },
+    {
+      "entropy": 0.12596421541646122,
+      "epoch": 7.535162950257289,
+      "grad_norm": 0.4448779225349426,
+      "learning_rate": 6.896340504795279e-05,
+      "loss": 0.09709668159484863,
+      "mean_token_accuracy": 0.9666328871250153,
+      "num_tokens": 6384710.0,
+      "step": 3300
+    },
+    {
+      "entropy": 0.1288480256870389,
+      "epoch": 7.649514008004574,
+      "grad_norm": 0.4604122042655945,
+      "learning_rate": 6.308718620568478e-05,
+      "loss": 0.10019060134887696,
+      "mean_token_accuracy": 0.9648687875270844,
+      "num_tokens": 6480406.0,
+      "step": 3350
+    },
+    {
+      "entropy": 0.11791575787588954,
+      "epoch": 7.763865065751858,
+      "grad_norm": 0.38850095868110657,
+      "learning_rate": 5.742479330312844e-05,
+      "loss": 0.09592323303222657,
+      "mean_token_accuracy": 0.9676644733548164,
+      "num_tokens": 6577776.0,
+      "step": 3400
+    },
+    {
+      "entropy": 0.12499945312738418,
+      "epoch": 7.878216123499143,
+      "grad_norm": 0.2908262014389038,
+      "learning_rate": 5.198521611996973e-05,
+      "loss": 0.09701567649841308,
+      "mean_token_accuracy": 0.9663982436060905,
+      "num_tokens": 6676978.0,
+      "step": 3450
+    },
+    {
+      "entropy": 0.12234330251812935,
+      "epoch": 7.992567181246427,
+      "grad_norm": 0.3459921181201935,
+      "learning_rate": 4.6777090687176785e-05,
+      "loss": 0.0960183334350586,
+      "mean_token_accuracy": 0.967001935839653,
+      "num_tokens": 6775723.0,
+      "step": 3500
+    },
+    {
+      "epoch": 8.0,
+      "eval_entropy": 0.23162500277487902,
+      "eval_loss": 0.9412578344345093,
+      "eval_mean_token_accuracy": 0.8381427713554271,
+      "eval_num_tokens": 6780848.0,
+      "eval_runtime": 81.7803,
+      "eval_samples_per_second": 13.328,
+      "eval_steps_per_second": 1.675,
+      "step": 3504
+    }
+  ],
+  "logging_steps": 50,
+  "max_steps": 4380,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 10,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 2.8173485110345114e+17,
+  "train_batch_size": 4,
+  "trial_name": null,
+  "trial_params": null
+}

systematicity_original_Swedish/Qwen3.5-4B-Base_systematicity_splits_original_features_train_systematicity_splits_original_features_test1/checkpoint-3942/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

systematicity_original_Swedish/Qwen3.5-4B-Base_systematicity_splits_original_features_train_systematicity_splits_original_features_test1/checkpoint-3942/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": 0.026477173062682283,
+  "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": [
+    "q_proj",
+    "k_proj",
+    "down_proj",
+    "v_proj",
+    "gate_proj",
+    "o_proj",
+    "up_proj"
+  ],
+  "target_parameters": null,
+  "task_type": "CAUSAL_LM",
+  "trainable_token_indices": null,
+  "use_dora": false,
+  "use_qalora": false,
+  "use_rslora": false
+}

systematicity_original_Swedish/Qwen3.5-4B-Base_systematicity_splits_original_features_train_systematicity_splits_original_features_test1/checkpoint-3942/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 %}

systematicity_original_Swedish/Qwen3.5-4B-Base_systematicity_splits_original_features_train_systematicity_splits_original_features_test1/checkpoint-3942/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|>"
+}

systematicity_original_Swedish/Qwen3.5-4B-Base_systematicity_splits_original_features_train_systematicity_splits_original_features_test1/checkpoint-3942/trainer_state.json

@@ -0,0 +1,913 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 9.0,
+  "eval_steps": 500,
+  "global_step": 3942,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "entropy": 1.4168990224599838,
+      "epoch": 0.11435105774728416,
+      "grad_norm": 1.3415299654006958,
+      "learning_rate": 4.424985043427545e-05,
+      "loss": 1.2160142517089845,
+      "mean_token_accuracy": 0.724469550549984,
+      "num_tokens": 96586.0,
+      "step": 50
+    },
+    {
+      "entropy": 0.7365697559714317,
+      "epoch": 0.22870211549456831,
+      "grad_norm": 0.9586130380630493,
+      "learning_rate": 8.940275904067897e-05,
+      "loss": 0.6445799255371094,
+      "mean_token_accuracy": 0.822787437736988,
+      "num_tokens": 190720.0,
+      "step": 100
+    },
+    {
+      "entropy": 0.691228820681572,
+      "epoch": 0.34305317324185247,
+      "grad_norm": 0.7875016331672668,
+      "learning_rate": 0.0001345556676470825,
+      "loss": 0.6011501312255859,
+      "mean_token_accuracy": 0.8325563898682594,
+      "num_tokens": 288515.0,
+      "step": 150
+    },
+    {
+      "entropy": 0.6706234656274319,
+      "epoch": 0.45740423098913663,
+      "grad_norm": 0.7867940068244934,
+      "learning_rate": 0.000179708576253486,
+      "loss": 0.5852980422973633,
+      "mean_token_accuracy": 0.83610415995121,
+      "num_tokens": 387874.0,
+      "step": 200
+    },
+    {
+      "entropy": 0.6558432649075985,
+      "epoch": 0.5717552887364208,
+      "grad_norm": 0.8049650192260742,
+      "learning_rate": 0.00022486148485988957,
+      "loss": 0.5714554977416992,
+      "mean_token_accuracy": 0.8380795314908027,
+      "num_tokens": 487865.0,
+      "step": 250
+    },
+    {
+      "entropy": 0.6612135827541351,
+      "epoch": 0.6861063464837049,
+      "grad_norm": 0.6776595711708069,
+      "learning_rate": 0.0002700143934662931,
+      "loss": 0.5815717315673828,
+      "mean_token_accuracy": 0.8377441626787185,
+      "num_tokens": 584269.0,
+      "step": 300
+    },
+    {
+      "entropy": 0.6668009972572326,
+      "epoch": 0.8004574042309891,
+      "grad_norm": 1.132367491722107,
+      "learning_rate": 0.0003151673020726966,
+      "loss": 0.5866946411132813,
+      "mean_token_accuracy": 0.8364784654974937,
+      "num_tokens": 676067.0,
+      "step": 350
+    },
+    {
+      "entropy": 0.6667388904094697,
+      "epoch": 0.9148084619782733,
+      "grad_norm": 0.9101859331130981,
+      "learning_rate": 0.0003603202106791001,
+      "loss": 0.5946685028076172,
+      "mean_token_accuracy": 0.8331978982686996,
+      "num_tokens": 774009.0,
+      "step": 400
+    },
+    {
+      "epoch": 1.0,
+      "eval_entropy": 0.5918735759536715,
+      "eval_loss": 0.6536311507225037,
+      "eval_mean_token_accuracy": 0.8276140002438622,
+      "eval_num_tokens": 847606.0,
+      "eval_runtime": 82.2575,
+      "eval_samples_per_second": 13.251,
+      "eval_steps_per_second": 1.666,
+      "step": 438
+    },
+    {
+      "entropy": 0.654371697890577,
+      "epoch": 1.0274442538593482,
+      "grad_norm": 0.923791229724884,
+      "learning_rate": 0.00039553187999846057,
+      "loss": 0.5879265594482422,
+      "mean_token_accuracy": 0.8356293612325252,
+      "num_tokens": 870214.0,
+      "step": 450
+    },
+    {
+      "entropy": 0.6167843751609325,
+      "epoch": 1.1417953116066324,
+      "grad_norm": 0.7943944334983826,
+      "learning_rate": 0.00039530582686207966,
+      "loss": 0.552466049194336,
+      "mean_token_accuracy": 0.8413123995065689,
+      "num_tokens": 971584.0,
+      "step": 500
+    },
+    {
+      "entropy": 0.6443017952144146,
+      "epoch": 1.2561463693539165,
+      "grad_norm": 1.018168568611145,
+      "learning_rate": 0.00039476615968757247,
+      "loss": 0.5773007583618164,
+      "mean_token_accuracy": 0.8377226310968399,
+      "num_tokens": 1063906.0,
+      "step": 550
+    },
+    {
+      "entropy": 0.6407185651361942,
+      "epoch": 1.3704974271012007,
+      "grad_norm": 0.9198986291885376,
+      "learning_rate": 0.00039391373526624385,
+      "loss": 0.572706184387207,
+      "mean_token_accuracy": 0.8367982304096222,
+      "num_tokens": 1159282.0,
+      "step": 600
+    },
+    {
+      "entropy": 0.6494526116549969,
+      "epoch": 1.4848484848484849,
+      "grad_norm": 1.0277528762817383,
+      "learning_rate": 0.00039274990693190134,
+      "loss": 0.578397102355957,
+      "mean_token_accuracy": 0.836915987432003,
+      "num_tokens": 1254998.0,
+      "step": 650
+    },
+    {
+      "entropy": 0.6396139825880528,
+      "epoch": 1.599199542595769,
+      "grad_norm": 0.8996608257293701,
+      "learning_rate": 0.0003912765224122631,
+      "loss": 0.5766008758544922,
+      "mean_token_accuracy": 0.8396538972854615,
+      "num_tokens": 1351877.0,
+      "step": 700
+    },
+    {
+      "entropy": 0.6245273813605309,
+      "epoch": 1.7135506003430532,
+      "grad_norm": 1.1999542713165283,
+      "learning_rate": 0.00038949592089545203,
+      "loss": 0.5718669509887695,
+      "mean_token_accuracy": 0.8406938418745995,
+      "num_tokens": 1448801.0,
+      "step": 750
+    },
+    {
+      "entropy": 0.6350119508802891,
+      "epoch": 1.8279016580903373,
+      "grad_norm": 0.9171736836433411,
+      "learning_rate": 0.00038741092931623213,
+      "loss": 0.5792041778564453,
+      "mean_token_accuracy": 0.8370065066218376,
+      "num_tokens": 1547040.0,
+      "step": 800
+    },
+    {
+      "entropy": 0.6106068694591522,
+      "epoch": 1.9422527158376215,
+      "grad_norm": 0.6869989037513733,
+      "learning_rate": 0.0003850248578678851,
+      "loss": 0.5748995590209961,
+      "mean_token_accuracy": 0.8383069199323654,
+      "num_tokens": 1646444.0,
+      "step": 850
+    },
+    {
+      "epoch": 2.0,
+      "eval_entropy": 0.5696834810893902,
+      "eval_loss": 0.6577129364013672,
+      "eval_mean_token_accuracy": 0.8307611911836332,
+      "eval_num_tokens": 1695212.0,
+      "eval_runtime": 81.951,
+      "eval_samples_per_second": 13.301,
+      "eval_steps_per_second": 1.672,
+      "step": 876
+    },
+    {
+      "entropy": 0.572038494088323,
+      "epoch": 2.0548885077186965,
+      "grad_norm": 0.7493197321891785,
+      "learning_rate": 0.00038234149474685104,
+      "loss": 0.5193290710449219,
+      "mean_token_accuracy": 0.8497188209277119,
+      "num_tokens": 1738782.0,
+      "step": 900
+    },
+    {
+      "entropy": 0.5478819286823273,
+      "epoch": 2.1692395654659804,
+      "grad_norm": 1.2549943923950195,
+      "learning_rate": 0.0003793651001384773,
+      "loss": 0.4793633270263672,
+      "mean_token_accuracy": 0.8575673401355743,
+      "num_tokens": 1837910.0,
+      "step": 950
+    },
+    {
+      "entropy": 0.5431128491461277,
+      "epoch": 2.283590623213265,
+      "grad_norm": 0.8713223338127136,
+      "learning_rate": 0.00037610039945342474,
+      "loss": 0.4849379348754883,
+      "mean_token_accuracy": 0.8562448808550834,
+      "num_tokens": 1937110.0,
+      "step": 1000
+    },
+    {
+      "entropy": 0.5498340383172036,
+      "epoch": 2.3979416809605487,
+      "grad_norm": 0.6592629551887512,
+      "learning_rate": 0.00037255257582546804,
+      "loss": 0.493680419921875,
+      "mean_token_accuracy": 0.854153895676136,
+      "num_tokens": 2036890.0,
+      "step": 1050
+    },
+    {
+      "entropy": 0.5554873237013817,
+      "epoch": 2.512292738707833,
+      "grad_norm": 0.7878748774528503,
+      "learning_rate": 0.0003687272618826019,
+      "loss": 0.4951488494873047,
+      "mean_token_accuracy": 0.8540353265404701,
+      "num_tokens": 2137239.0,
+      "step": 1100
+    },
+    {
+      "entropy": 0.552360344529152,
+      "epoch": 2.6266437964551175,
+      "grad_norm": 0.7380415797233582,
+      "learning_rate": 0.00036463053080451683,
+      "loss": 0.49989810943603513,
+      "mean_token_accuracy": 0.8533406674861908,
+      "num_tokens": 2231081.0,
+      "step": 1150
+    },
+    {
+      "entropy": 0.5345852719247342,
+      "epoch": 2.7409948542024014,
+      "grad_norm": 0.7600613832473755,
+      "learning_rate": 0.0003602688866806424,
+      "loss": 0.5011516189575196,
+      "mean_token_accuracy": 0.8527707865834236,
+      "num_tokens": 2325503.0,
+      "step": 1200
+    },
+    {
+      "entropy": 0.532319468408823,
+      "epoch": 2.8553459119496853,
+      "grad_norm": 0.8197540640830994,
+      "learning_rate": 0.000355649254184066,
+      "loss": 0.5023023605346679,
+      "mean_token_accuracy": 0.8530397671461105,
+      "num_tokens": 2420627.0,
+      "step": 1250
+    },
+    {
+      "entropy": 0.5198138700425625,
+      "epoch": 2.9696969696969697,
+      "grad_norm": 0.8660516142845154,
+      "learning_rate": 0.00035077896757771987,
+      "loss": 0.4941487121582031,
+      "mean_token_accuracy": 0.855004439651966,
+      "num_tokens": 2518051.0,
+      "step": 1300
+    },
+    {
+      "epoch": 3.0,
+      "eval_entropy": 0.5642598084289662,
+      "eval_loss": 0.6125267744064331,
+      "eval_mean_token_accuracy": 0.8375998204641969,
+      "eval_num_tokens": 2542818.0,
+      "eval_runtime": 81.5524,
+      "eval_samples_per_second": 13.366,
+      "eval_steps_per_second": 1.68,
+      "step": 1314
+    },
+    {
+      "entropy": 0.45835615868495805,
+      "epoch": 3.0823327615780447,
+      "grad_norm": 0.734519362449646,
+      "learning_rate": 0.00034566575907029245,
+      "loss": 0.4334117889404297,
+      "mean_token_accuracy": 0.8698135275526095,
+      "num_tokens": 2613861.0,
+      "step": 1350
+    },
+    {
+      "entropy": 0.44648877054452896,
+      "epoch": 3.1966838193253286,
+      "grad_norm": 0.687981128692627,
+      "learning_rate": 0.00034031774654034867,
+      "loss": 0.40152584075927733,
+      "mean_token_accuracy": 0.8759277358651161,
+      "num_tokens": 2709451.0,
+      "step": 1400
+    },
+    {
+      "entropy": 0.4593352344632149,
+      "epoch": 3.311034877072613,
+      "grad_norm": 0.6310222148895264,
+      "learning_rate": 0.0003347434206481497,
+      "loss": 0.41885005950927734,
+      "mean_token_accuracy": 0.8714999267458916,
+      "num_tokens": 2801336.0,
+      "step": 1450
+    },
+    {
+      "entropy": 0.46372457168996334,
+      "epoch": 3.425385934819897,
+      "grad_norm": 0.6795068979263306,
+      "learning_rate": 0.0003289516313556339,
+      "loss": 0.4194014358520508,
+      "mean_token_accuracy": 0.8709167090058326,
+      "num_tokens": 2898945.0,
+      "step": 1500
+    },
+    {
+      "entropy": 0.4548304344713688,
+      "epoch": 3.5397369925671813,
+      "grad_norm": 0.9671820402145386,
+      "learning_rate": 0.00032295157387595997,
+      "loss": 0.4178035736083984,
+      "mean_token_accuracy": 0.8710319435596466,
+      "num_tokens": 2993378.0,
+      "step": 1550
+    },
+    {
+      "entropy": 0.4532127542793751,
+      "epoch": 3.6540880503144653,
+      "grad_norm": 0.7656959295272827,
+      "learning_rate": 0.0003167527740749187,
+      "loss": 0.42567527770996094,
+      "mean_token_accuracy": 0.869033499956131,
+      "num_tokens": 3091323.0,
+      "step": 1600
+    },
+    {
+      "entropy": 0.4460975781828165,
+      "epoch": 3.7684391080617496,
+      "grad_norm": 0.7811372876167297,
+      "learning_rate": 0.0003103650733473916,
+      "loss": 0.4148756790161133,
+      "mean_token_accuracy": 0.8709280017018318,
+      "num_tokens": 3193194.0,
+      "step": 1650
+    },
+    {
+      "entropy": 0.47125090911984446,
+      "epoch": 3.8827901658090336,
+      "grad_norm": 0.8664485216140747,
+      "learning_rate": 0.00030379861299286587,
+      "loss": 0.4209564590454102,
+      "mean_token_accuracy": 0.8695067319273949,
+      "num_tokens": 3292840.0,
+      "step": 1700
+    },
+    {
+      "entropy": 0.4641740974783897,
+      "epoch": 3.997141223556318,
+      "grad_norm": 0.85616534948349,
+      "learning_rate": 0.00029706381811481263,
+      "loss": 0.4319478988647461,
+      "mean_token_accuracy": 0.8689105224609375,
+      "num_tokens": 3387511.0,
+      "step": 1750
+    },
+    {
+      "epoch": 4.0,
+      "eval_entropy": 0.4623802898138979,
+      "eval_loss": 0.6707634925842285,
+      "eval_mean_token_accuracy": 0.8327476525828786,
+      "eval_num_tokens": 3390424.0,
+      "eval_runtime": 81.9106,
+      "eval_samples_per_second": 13.307,
+      "eval_steps_per_second": 1.673,
+      "step": 1752
+    },
+    {
+      "entropy": 0.33750920285125674,
+      "epoch": 4.109777015437393,
+      "grad_norm": 0.6137083172798157,
+      "learning_rate": 0.00029017138106948915,
+      "loss": 0.31130544662475584,
+      "mean_token_accuracy": 0.8997744660692166,
+      "num_tokens": 3483795.0,
+      "step": 1800
+    },
+    {
+      "entropy": 0.3504758469760418,
+      "epoch": 4.224128073184677,
+      "grad_norm": 0.6796069741249084,
+      "learning_rate": 0.00028313224449044337,
+      "loss": 0.31956762313842774,
+      "mean_token_accuracy": 0.895492787361145,
+      "num_tokens": 3576928.0,
+      "step": 1850
+    },
+    {
+      "entropy": 0.34297160781919955,
+      "epoch": 4.338479130931961,
+      "grad_norm": 0.7507141828536987,
+      "learning_rate": 0.00027595758391566995,
+      "loss": 0.31722068786621094,
+      "mean_token_accuracy": 0.8957494455575943,
+      "num_tokens": 3674125.0,
+      "step": 1900
+    },
+    {
+      "entropy": 0.3572557730227709,
+      "epoch": 4.452830188679245,
+      "grad_norm": 0.7509586215019226,
+      "learning_rate": 0.00026865879004500166,
+      "loss": 0.32728019714355466,
+      "mean_token_accuracy": 0.894504280090332,
+      "num_tokens": 3769853.0,
+      "step": 1950
+    },
+    {
+      "entropy": 0.35588217943906786,
+      "epoch": 4.56718124642653,
+      "grad_norm": 0.6168456077575684,
+      "learning_rate": 0.0002612474506559027,
+      "loss": 0.3304343795776367,
+      "mean_token_accuracy": 0.8935999310016632,
+      "num_tokens": 3867574.0,
+      "step": 2000
+    },
+    {
+      "entropy": 0.3554471937566996,
+      "epoch": 4.681532304173814,
+      "grad_norm": 0.6922969222068787,
+      "learning_rate": 0.00025373533220637623,
+      "loss": 0.3315126037597656,
+      "mean_token_accuracy": 0.8925720381736756,
+      "num_tokens": 3969484.0,
+      "step": 2050
+    },
+    {
+      "entropy": 0.35552985966205597,
+      "epoch": 4.795883361921097,
+      "grad_norm": 0.6013802289962769,
+      "learning_rate": 0.00024613436115419326,
+      "loss": 0.3315456771850586,
+      "mean_token_accuracy": 0.8929180511832238,
+      "num_tokens": 4063480.0,
+      "step": 2100
+    },
+    {
+      "entropy": 0.3478537338972092,
+      "epoch": 4.910234419668382,
+      "grad_norm": 0.8090653419494629,
+      "learning_rate": 0.00023845660502210138,
+      "loss": 0.32579368591308594,
+      "mean_token_accuracy": 0.893961855173111,
+      "num_tokens": 4162925.0,
+      "step": 2150
+    },
+    {
+      "epoch": 5.0,
+      "eval_entropy": 0.38475005339532004,
+      "eval_loss": 0.6992377042770386,
+      "eval_mean_token_accuracy": 0.8384414993933518,
+      "eval_num_tokens": 4238030.0,
+      "eval_runtime": 81.1991,
+      "eval_samples_per_second": 13.424,
+      "eval_steps_per_second": 1.687,
+      "step": 2190
+    },
+    {
+      "entropy": 0.3381291570397198,
+      "epoch": 5.022870211549457,
+      "grad_norm": 0.7052969932556152,
+      "learning_rate": 0.00023071425323907478,
+      "loss": 0.3124395370483398,
+      "mean_token_accuracy": 0.8997272504162668,
+      "num_tokens": 4257610.0,
+      "step": 2200
+    },
+    {
+      "entropy": 0.2494013351202011,
+      "epoch": 5.137221269296741,
+      "grad_norm": 0.8653550148010254,
+      "learning_rate": 0.0002229195977880219,
+      "loss": 0.21398515701293946,
+      "mean_token_accuracy": 0.9277590665221215,
+      "num_tokens": 4353703.0,
+      "step": 2250
+    },
+    {
+      "entropy": 0.27507017262279987,
+      "epoch": 5.251572327044025,
+      "grad_norm": 0.7887904644012451,
+      "learning_rate": 0.00021508501369067626,
+      "loss": 0.22058496475219727,
+      "mean_token_accuracy": 0.9253441894054413,
+      "num_tokens": 4449829.0,
+      "step": 2300
+    },
+    {
+      "entropy": 0.2582047886401415,
+      "epoch": 5.3659233847913095,
+      "grad_norm": 0.8728304505348206,
+      "learning_rate": 0.0002072229393606514,
+      "loss": 0.22437936782836915,
+      "mean_token_accuracy": 0.9235965391993523,
+      "num_tokens": 4545644.0,
+      "step": 2350
+    },
+    {
+      "entropy": 0.26232284381985665,
+      "epoch": 5.480274442538594,
+      "grad_norm": 0.6688716411590576,
+      "learning_rate": 0.00019934585685585428,
+      "loss": 0.23260833740234374,
+      "mean_token_accuracy": 0.9217587423324585,
+      "num_tokens": 4637941.0,
+      "step": 2400
+    },
+    {
+      "entropy": 0.2619731902331114,
+      "epoch": 5.594625500285877,
+      "grad_norm": 0.597118616104126,
+      "learning_rate": 0.00019146627206160645,
+      "loss": 0.22075172424316405,
+      "mean_token_accuracy": 0.9252614346146584,
+      "num_tokens": 4738175.0,
+      "step": 2450
+    },
+    {
+      "entropy": 0.26591138120740654,
+      "epoch": 5.708976558033162,
+      "grad_norm": 0.7602102160453796,
+      "learning_rate": 0.00018359669483593703,
+      "loss": 0.22200309753417968,
+      "mean_token_accuracy": 0.9250017151236534,
+      "num_tokens": 4837766.0,
+      "step": 2500
+    },
+    {
+      "entropy": 0.2587306099012494,
+      "epoch": 5.823327615780446,
+      "grad_norm": 0.6875442862510681,
+      "learning_rate": 0.00017574961914856833,
+      "loss": 0.22627996444702148,
+      "mean_token_accuracy": 0.9234379661083222,
+      "num_tokens": 4935670.0,
+      "step": 2550
+    },
+    {
+      "entropy": 0.2689222282171249,
+      "epoch": 5.9376786735277305,
+      "grad_norm": 0.654289186000824,
+      "learning_rate": 0.00016793750324512556,
+      "loss": 0.23186542510986327,
+      "mean_token_accuracy": 0.9214966455101967,
+      "num_tokens": 5030799.0,
+      "step": 2600
+    },
+    {
+      "epoch": 6.0,
+      "eval_entropy": 0.34485617432281046,
+      "eval_loss": 0.7253938317298889,
+      "eval_mean_token_accuracy": 0.8377144193127207,
+      "eval_num_tokens": 5085636.0,
+      "eval_runtime": 81.5281,
+      "eval_samples_per_second": 13.37,
+      "eval_steps_per_second": 1.68,
+      "step": 2628
+    },
+    {
+      "entropy": 0.22255296551334072,
+      "epoch": 6.050314465408805,
+      "grad_norm": 0.5370154976844788,
+      "learning_rate": 0.00016017274986806444,
+      "loss": 0.18560497283935548,
+      "mean_token_accuracy": 0.9371391822238864,
+      "num_tokens": 5126968.0,
+      "step": 2650
+    },
+    {
+      "entropy": 0.1745598777383566,
+      "epoch": 6.164665523156089,
+      "grad_norm": 0.6606088280677795,
+      "learning_rate": 0.00015246768656571565,
+      "loss": 0.14104312896728516,
+      "mean_token_accuracy": 0.9522112870216369,
+      "num_tokens": 5222926.0,
+      "step": 2700
+    },
+    {
+      "entropy": 0.16993012264370919,
+      "epoch": 6.279016580903374,
+      "grad_norm": 0.6963379383087158,
+      "learning_rate": 0.00014483454612071133,
+      "loss": 0.14232961654663087,
+      "mean_token_accuracy": 0.9532496619224549,
+      "num_tokens": 5317743.0,
+      "step": 2750
+    },
+    {
+      "entropy": 0.16726660918444394,
+      "epoch": 6.393367638650657,
+      "grad_norm": 0.7356508374214172,
+      "learning_rate": 0.00013728544712886273,
+      "loss": 0.13951081275939942,
+      "mean_token_accuracy": 0.9533490630984306,
+      "num_tokens": 5415584.0,
+      "step": 2800
+    },
+    {
+      "entropy": 0.17247022703289985,
+      "epoch": 6.507718696397942,
+      "grad_norm": 0.6522324681282043,
+      "learning_rate": 0.00012983237475932584,
+      "loss": 0.13756629943847656,
+      "mean_token_accuracy": 0.9529287284612655,
+      "num_tokens": 5516998.0,
+      "step": 2850
+    },
+    {
+      "entropy": 0.1730361183360219,
+      "epoch": 6.622069754145226,
+      "grad_norm": 0.747295618057251,
+      "learning_rate": 0.00012248716172659724,
+      "loss": 0.1429327392578125,
+      "mean_token_accuracy": 0.9519825592637062,
+      "num_tokens": 5614424.0,
+      "step": 2900
+    },
+    {
+      "entropy": 0.17447776831686496,
+      "epoch": 6.7364208118925095,
+      "grad_norm": 0.624346137046814,
+      "learning_rate": 0.00011526146950455331,
+      "loss": 0.14175209999084473,
+      "mean_token_accuracy": 0.9515316143631936,
+      "num_tokens": 5712047.0,
+      "step": 2950
+    },
+    {
+      "entropy": 0.181520360969007,
+      "epoch": 6.850771869639794,
+      "grad_norm": 0.747399091720581,
+      "learning_rate": 0.00010816676981235426,
+      "loss": 0.14949700355529785,
+      "mean_token_accuracy": 0.9500517123937606,
+      "num_tokens": 5805524.0,
+      "step": 3000
+    },
+    {
+      "entropy": 0.17918075375258924,
+      "epoch": 6.965122927387078,
+      "grad_norm": 0.55885249376297,
+      "learning_rate": 0.00010121432640160985,
+      "loss": 0.14284159660339354,
+      "mean_token_accuracy": 0.9515545222163201,
+      "num_tokens": 5902110.0,
+      "step": 3050
+    },
+    {
+      "epoch": 7.0,
+      "eval_entropy": 0.2829361812476694,
+      "eval_loss": 0.8297234177589417,
+      "eval_mean_token_accuracy": 0.8362262305552072,
+      "eval_num_tokens": 5933242.0,
+      "eval_runtime": 81.6661,
+      "eval_samples_per_second": 13.347,
+      "eval_steps_per_second": 1.678,
+      "step": 3066
+    },
+    {
+      "entropy": 0.13936093423118445,
+      "epoch": 7.077758719268153,
+      "grad_norm": 0.6394845843315125,
+      "learning_rate": 9.441517717371896e-05,
+      "loss": 0.11461298942565917,
+      "mean_token_accuracy": 0.9617519581378414,
+      "num_tokens": 5994998.0,
+      "step": 3100
+    },
+    {
+      "entropy": 0.12445554660633207,
+      "epoch": 7.192109777015437,
+      "grad_norm": 0.3524336814880371,
+      "learning_rate": 8.778011665577682e-05,
+      "loss": 0.09778708457946778,
+      "mean_token_accuracy": 0.9672333124279976,
+      "num_tokens": 6089591.0,
+      "step": 3150
+    },
+    {
+      "entropy": 0.11911977402865886,
+      "epoch": 7.306460834762722,
+      "grad_norm": 0.5336443185806274,
+      "learning_rate": 8.131967886286935e-05,
+      "loss": 0.0957869529724121,
+      "mean_token_accuracy": 0.9681107413768768,
+      "num_tokens": 6189134.0,
+      "step": 3200
+    },
+    {
+      "entropy": 0.12348919529467821,
+      "epoch": 7.420811892510006,
+      "grad_norm": 0.3701528310775757,
+      "learning_rate": 7.504412057396483e-05,
+      "loss": 0.09508245468139648,
+      "mean_token_accuracy": 0.9655187261104584,
+      "num_tokens": 6290034.0,
+      "step": 3250
+    },
+    {
+      "entropy": 0.12596421541646122,
+      "epoch": 7.535162950257289,
+      "grad_norm": 0.4448779225349426,
+      "learning_rate": 6.896340504795279e-05,
+      "loss": 0.09709668159484863,
+      "mean_token_accuracy": 0.9666328871250153,
+      "num_tokens": 6384710.0,
+      "step": 3300
+    },
+    {
+      "entropy": 0.1288480256870389,
+      "epoch": 7.649514008004574,
+      "grad_norm": 0.4604122042655945,
+      "learning_rate": 6.308718620568478e-05,
+      "loss": 0.10019060134887696,
+      "mean_token_accuracy": 0.9648687875270844,
+      "num_tokens": 6480406.0,
+      "step": 3350
+    },
+    {
+      "entropy": 0.11791575787588954,
+      "epoch": 7.763865065751858,
+      "grad_norm": 0.38850095868110657,
+      "learning_rate": 5.742479330312844e-05,
+      "loss": 0.09592323303222657,
+      "mean_token_accuracy": 0.9676644733548164,
+      "num_tokens": 6577776.0,
+      "step": 3400
+    },
+    {
+      "entropy": 0.12499945312738418,
+      "epoch": 7.878216123499143,
+      "grad_norm": 0.2908262014389038,
+      "learning_rate": 5.198521611996973e-05,
+      "loss": 0.09701567649841308,
+      "mean_token_accuracy": 0.9663982436060905,
+      "num_tokens": 6676978.0,
+      "step": 3450
+    },
+    {
+      "entropy": 0.12234330251812935,
+      "epoch": 7.992567181246427,
+      "grad_norm": 0.3459921181201935,
+      "learning_rate": 4.6777090687176785e-05,
+      "loss": 0.0960183334350586,
+      "mean_token_accuracy": 0.967001935839653,
+      "num_tokens": 6775723.0,
+      "step": 3500
+    },
+    {
+      "epoch": 8.0,
+      "eval_entropy": 0.23162500277487902,
+      "eval_loss": 0.9412578344345093,
+      "eval_mean_token_accuracy": 0.8381427713554271,
+      "eval_num_tokens": 6780848.0,
+      "eval_runtime": 81.7803,
+      "eval_samples_per_second": 13.328,
+      "eval_steps_per_second": 1.675,
+      "step": 3504
+    },
+    {
+      "entropy": 0.10137141654787934,
+      "epoch": 8.1052029731275,
+      "grad_norm": 0.26042085886001587,
+      "learning_rate": 4.1808685576186306e-05,
+      "loss": 0.07917721271514892,
+      "mean_token_accuracy": 0.9726437660038169,
+      "num_tokens": 6873669.0,
+      "step": 3550
+    },
+    {
+      "entropy": 0.10200918957591057,
+      "epoch": 8.219554030874786,
+      "grad_norm": 0.24989385902881622,
+      "learning_rate": 3.708788877147874e-05,
+      "loss": 0.07984754085540771,
+      "mean_token_accuracy": 0.9713884252309799,
+      "num_tokens": 6969306.0,
+      "step": 3600
+    },
+    {
+      "entropy": 0.09767990596592427,
+      "epoch": 8.33390508862207,
+      "grad_norm": 0.3814045488834381,
+      "learning_rate": 3.2622195147384906e-05,
+      "loss": 0.07905120849609375,
+      "mean_token_accuracy": 0.9719935733079911,
+      "num_tokens": 7067472.0,
+      "step": 3650
+    },
+    {
+      "entropy": 0.10590592253953218,
+      "epoch": 8.448256146369355,
+      "grad_norm": 0.296535462141037,
+      "learning_rate": 2.8418694569005194e-05,
+      "loss": 0.08534525871276856,
+      "mean_token_accuracy": 0.9699256777763366,
+      "num_tokens": 7160953.0,
+      "step": 3700
+    },
+    {
+      "entropy": 0.10513897916302085,
+      "epoch": 8.562607204116638,
+      "grad_norm": 0.24547256529331207,
+      "learning_rate": 2.448406063613386e-05,
+      "loss": 0.07970548152923584,
+      "mean_token_accuracy": 0.9707170671224594,
+      "num_tokens": 7257979.0,
+      "step": 3750
+    },
+    {
+      "entropy": 0.10294155770912766,
+      "epoch": 8.676958261863922,
+      "grad_norm": 0.195679172873497,
+      "learning_rate": 2.082454008805752e-05,
+      "loss": 0.08084090232849121,
+      "mean_token_accuracy": 0.9708512514829636,
+      "num_tokens": 7355068.0,
+      "step": 3800
+    },
+    {
+      "entropy": 0.10045578340068459,
+      "epoch": 8.791309319611207,
+      "grad_norm": 0.21481232345104218,
+      "learning_rate": 1.7445942886050436e-05,
+      "loss": 0.0806255054473877,
+      "mean_token_accuracy": 0.971126164495945,
+      "num_tokens": 7451068.0,
+      "step": 3850
+    },
+    {
+      "entropy": 0.1046586587280035,
+      "epoch": 8.90566037735849,
+      "grad_norm": 0.34530165791511536,
+      "learning_rate": 1.4353632989310662e-05,
+      "loss": 0.08288439750671386,
+      "mean_token_accuracy": 0.9699718147516251,
+      "num_tokens": 7548327.0,
+      "step": 3900
+    },
+    {
+      "epoch": 9.0,
+      "eval_entropy": 0.2088741992946959,
+      "eval_loss": 1.049438238143921,
+      "eval_mean_token_accuracy": 0.8382186280549878,
+      "eval_num_tokens": 7628454.0,
+      "eval_runtime": 81.7324,
+      "eval_samples_per_second": 13.336,
+      "eval_steps_per_second": 1.676,
+      "step": 3942
+    }
+  ],
+  "logging_steps": 50,
+  "max_steps": 4380,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 10,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 3.1683226343807386e+17,
+  "train_batch_size": 4,
+  "trial_name": null,
+  "trial_params": null
+}

systematicity_original_Swedish/Qwen3.5-4B-Base_systematicity_splits_original_features_train_systematicity_splits_original_features_test1/checkpoint-438/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

systematicity_original_Swedish/Qwen3.5-4B-Base_systematicity_splits_original_features_train_systematicity_splits_original_features_test1/checkpoint-438/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": 0.026477173062682283,
+  "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": [
+    "q_proj",
+    "k_proj",
+    "down_proj",
+    "v_proj",
+    "gate_proj",
+    "o_proj",
+    "up_proj"
+  ],
+  "target_parameters": null,
+  "task_type": "CAUSAL_LM",
+  "trainable_token_indices": null,
+  "use_dora": false,
+  "use_qalora": false,
+  "use_rslora": false
+}

systematicity_original_Swedish/Qwen3.5-4B-Base_systematicity_splits_original_features_train_systematicity_splits_original_features_test1/checkpoint-438/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 %}

systematicity_original_Swedish/Qwen3.5-4B-Base_systematicity_splits_original_features_train_systematicity_splits_original_features_test1/checkpoint-438/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|>"
+}

systematicity_original_Swedish/Qwen3.5-4B-Base_systematicity_splits_original_features_train_systematicity_splits_original_features_test1/checkpoint-438/trainer_state.json

@@ -0,0 +1,125 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.0,
+  "eval_steps": 500,
+  "global_step": 438,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "entropy": 1.4168990224599838,
+      "epoch": 0.11435105774728416,
+      "grad_norm": 1.3415299654006958,
+      "learning_rate": 4.424985043427545e-05,
+      "loss": 1.2160142517089845,
+      "mean_token_accuracy": 0.724469550549984,
+      "num_tokens": 96586.0,
+      "step": 50
+    },
+    {
+      "entropy": 0.7365697559714317,
+      "epoch": 0.22870211549456831,
+      "grad_norm": 0.9586130380630493,
+      "learning_rate": 8.940275904067897e-05,
+      "loss": 0.6445799255371094,
+      "mean_token_accuracy": 0.822787437736988,
+      "num_tokens": 190720.0,
+      "step": 100
+    },
+    {
+      "entropy": 0.691228820681572,
+      "epoch": 0.34305317324185247,
+      "grad_norm": 0.7875016331672668,
+      "learning_rate": 0.0001345556676470825,
+      "loss": 0.6011501312255859,
+      "mean_token_accuracy": 0.8325563898682594,
+      "num_tokens": 288515.0,
+      "step": 150
+    },
+    {
+      "entropy": 0.6706234656274319,
+      "epoch": 0.45740423098913663,
+      "grad_norm": 0.7867940068244934,
+      "learning_rate": 0.000179708576253486,
+      "loss": 0.5852980422973633,
+      "mean_token_accuracy": 0.83610415995121,
+      "num_tokens": 387874.0,
+      "step": 200
+    },
+    {
+      "entropy": 0.6558432649075985,
+      "epoch": 0.5717552887364208,
+      "grad_norm": 0.8049650192260742,
+      "learning_rate": 0.00022486148485988957,
+      "loss": 0.5714554977416992,
+      "mean_token_accuracy": 0.8380795314908027,
+      "num_tokens": 487865.0,
+      "step": 250
+    },
+    {
+      "entropy": 0.6612135827541351,
+      "epoch": 0.6861063464837049,
+      "grad_norm": 0.6776595711708069,
+      "learning_rate": 0.0002700143934662931,
+      "loss": 0.5815717315673828,
+      "mean_token_accuracy": 0.8377441626787185,
+      "num_tokens": 584269.0,
+      "step": 300
+    },
+    {
+      "entropy": 0.6668009972572326,
+      "epoch": 0.8004574042309891,
+      "grad_norm": 1.132367491722107,
+      "learning_rate": 0.0003151673020726966,
+      "loss": 0.5866946411132813,
+      "mean_token_accuracy": 0.8364784654974937,
+      "num_tokens": 676067.0,
+      "step": 350
+    },
+    {
+      "entropy": 0.6667388904094697,
+      "epoch": 0.9148084619782733,
+      "grad_norm": 0.9101859331130981,
+      "learning_rate": 0.0003603202106791001,
+      "loss": 0.5946685028076172,
+      "mean_token_accuracy": 0.8331978982686996,
+      "num_tokens": 774009.0,
+      "step": 400
+    },
+    {
+      "epoch": 1.0,
+      "eval_entropy": 0.5918735759536715,
+      "eval_loss": 0.6536311507225037,
+      "eval_mean_token_accuracy": 0.8276140002438622,
+      "eval_num_tokens": 847606.0,
+      "eval_runtime": 82.2575,
+      "eval_samples_per_second": 13.251,
+      "eval_steps_per_second": 1.666,
+      "step": 438
+    }
+  ],
+  "logging_steps": 50,
+  "max_steps": 4380,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 10,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 3.519699610366771e+16,
+  "train_batch_size": 4,
+  "trial_name": null,
+  "trial_params": null
+}

systematicity_original_Swedish/Qwen3.5-4B-Base_systematicity_splits_original_features_train_systematicity_splits_original_features_test1/checkpoint-4380/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

systematicity_original_Swedish/Qwen3.5-4B-Base_systematicity_splits_original_features_train_systematicity_splits_original_features_test1/checkpoint-4380/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": 0.026477173062682283,
+  "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": [
+    "q_proj",
+    "k_proj",
+    "down_proj",
+    "v_proj",
+    "gate_proj",
+    "o_proj",
+    "up_proj"
+  ],
+  "target_parameters": null,
+  "task_type": "CAUSAL_LM",
+  "trainable_token_indices": null,
+  "use_dora": false,
+  "use_qalora": false,
+  "use_rslora": false
+}

systematicity_original_Swedish/Qwen3.5-4B-Base_systematicity_splits_original_features_train_systematicity_splits_original_features_test1/checkpoint-4380/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 %}

systematicity_original_Swedish/Qwen3.5-4B-Base_systematicity_splits_original_features_train_systematicity_splits_original_features_test1/checkpoint-4380/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|>"
+}

systematicity_original_Swedish/Qwen3.5-4B-Base_systematicity_splits_original_features_train_systematicity_splits_original_features_test1/checkpoint-4380/trainer_state.json

@@ -0,0 +1,1014 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 10.0,
+  "eval_steps": 500,
+  "global_step": 4380,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "entropy": 1.4168990224599838,
+      "epoch": 0.11435105774728416,
+      "grad_norm": 1.3415299654006958,
+      "learning_rate": 4.424985043427545e-05,
+      "loss": 1.2160142517089845,
+      "mean_token_accuracy": 0.724469550549984,
+      "num_tokens": 96586.0,
+      "step": 50
+    },
+    {
+      "entropy": 0.7365697559714317,
+      "epoch": 0.22870211549456831,
+      "grad_norm": 0.9586130380630493,
+      "learning_rate": 8.940275904067897e-05,
+      "loss": 0.6445799255371094,
+      "mean_token_accuracy": 0.822787437736988,
+      "num_tokens": 190720.0,
+      "step": 100
+    },
+    {
+      "entropy": 0.691228820681572,
+      "epoch": 0.34305317324185247,
+      "grad_norm": 0.7875016331672668,
+      "learning_rate": 0.0001345556676470825,
+      "loss": 0.6011501312255859,
+      "mean_token_accuracy": 0.8325563898682594,
+      "num_tokens": 288515.0,
+      "step": 150
+    },
+    {
+      "entropy": 0.6706234656274319,
+      "epoch": 0.45740423098913663,
+      "grad_norm": 0.7867940068244934,
+      "learning_rate": 0.000179708576253486,
+      "loss": 0.5852980422973633,
+      "mean_token_accuracy": 0.83610415995121,
+      "num_tokens": 387874.0,
+      "step": 200
+    },
+    {
+      "entropy": 0.6558432649075985,
+      "epoch": 0.5717552887364208,
+      "grad_norm": 0.8049650192260742,
+      "learning_rate": 0.00022486148485988957,
+      "loss": 0.5714554977416992,
+      "mean_token_accuracy": 0.8380795314908027,
+      "num_tokens": 487865.0,
+      "step": 250
+    },
+    {
+      "entropy": 0.6612135827541351,
+      "epoch": 0.6861063464837049,
+      "grad_norm": 0.6776595711708069,
+      "learning_rate": 0.0002700143934662931,
+      "loss": 0.5815717315673828,
+      "mean_token_accuracy": 0.8377441626787185,
+      "num_tokens": 584269.0,
+      "step": 300
+    },
+    {
+      "entropy": 0.6668009972572326,
+      "epoch": 0.8004574042309891,
+      "grad_norm": 1.132367491722107,
+      "learning_rate": 0.0003151673020726966,
+      "loss": 0.5866946411132813,
+      "mean_token_accuracy": 0.8364784654974937,
+      "num_tokens": 676067.0,
+      "step": 350
+    },
+    {
+      "entropy": 0.6667388904094697,
+      "epoch": 0.9148084619782733,
+      "grad_norm": 0.9101859331130981,
+      "learning_rate": 0.0003603202106791001,
+      "loss": 0.5946685028076172,
+      "mean_token_accuracy": 0.8331978982686996,
+      "num_tokens": 774009.0,
+      "step": 400
+    },
+    {
+      "epoch": 1.0,
+      "eval_entropy": 0.5918735759536715,
+      "eval_loss": 0.6536311507225037,
+      "eval_mean_token_accuracy": 0.8276140002438622,
+      "eval_num_tokens": 847606.0,
+      "eval_runtime": 82.2575,
+      "eval_samples_per_second": 13.251,
+      "eval_steps_per_second": 1.666,
+      "step": 438
+    },
+    {
+      "entropy": 0.654371697890577,
+      "epoch": 1.0274442538593482,
+      "grad_norm": 0.923791229724884,
+      "learning_rate": 0.00039553187999846057,
+      "loss": 0.5879265594482422,
+      "mean_token_accuracy": 0.8356293612325252,
+      "num_tokens": 870214.0,
+      "step": 450
+    },
+    {
+      "entropy": 0.6167843751609325,
+      "epoch": 1.1417953116066324,
+      "grad_norm": 0.7943944334983826,
+      "learning_rate": 0.00039530582686207966,
+      "loss": 0.552466049194336,
+      "mean_token_accuracy": 0.8413123995065689,
+      "num_tokens": 971584.0,
+      "step": 500
+    },
+    {
+      "entropy": 0.6443017952144146,
+      "epoch": 1.2561463693539165,
+      "grad_norm": 1.018168568611145,
+      "learning_rate": 0.00039476615968757247,
+      "loss": 0.5773007583618164,
+      "mean_token_accuracy": 0.8377226310968399,
+      "num_tokens": 1063906.0,
+      "step": 550
+    },
+    {
+      "entropy": 0.6407185651361942,
+      "epoch": 1.3704974271012007,
+      "grad_norm": 0.9198986291885376,
+      "learning_rate": 0.00039391373526624385,
+      "loss": 0.572706184387207,
+      "mean_token_accuracy": 0.8367982304096222,
+      "num_tokens": 1159282.0,
+      "step": 600
+    },
+    {
+      "entropy": 0.6494526116549969,
+      "epoch": 1.4848484848484849,
+      "grad_norm": 1.0277528762817383,
+      "learning_rate": 0.00039274990693190134,
+      "loss": 0.578397102355957,
+      "mean_token_accuracy": 0.836915987432003,
+      "num_tokens": 1254998.0,
+      "step": 650
+    },
+    {
+      "entropy": 0.6396139825880528,
+      "epoch": 1.599199542595769,
+      "grad_norm": 0.8996608257293701,
+      "learning_rate": 0.0003912765224122631,
+      "loss": 0.5766008758544922,
+      "mean_token_accuracy": 0.8396538972854615,
+      "num_tokens": 1351877.0,
+      "step": 700
+    },
+    {
+      "entropy": 0.6245273813605309,
+      "epoch": 1.7135506003430532,
+      "grad_norm": 1.1999542713165283,
+      "learning_rate": 0.00038949592089545203,
+      "loss": 0.5718669509887695,
+      "mean_token_accuracy": 0.8406938418745995,
+      "num_tokens": 1448801.0,
+      "step": 750
+    },
+    {
+      "entropy": 0.6350119508802891,
+      "epoch": 1.8279016580903373,
+      "grad_norm": 0.9171736836433411,
+      "learning_rate": 0.00038741092931623213,
+      "loss": 0.5792041778564453,
+      "mean_token_accuracy": 0.8370065066218376,
+      "num_tokens": 1547040.0,
+      "step": 800
+    },
+    {
+      "entropy": 0.6106068694591522,
+      "epoch": 1.9422527158376215,
+      "grad_norm": 0.6869989037513733,
+      "learning_rate": 0.0003850248578678851,
+      "loss": 0.5748995590209961,
+      "mean_token_accuracy": 0.8383069199323654,
+      "num_tokens": 1646444.0,
+      "step": 850
+    },
+    {
+      "epoch": 2.0,
+      "eval_entropy": 0.5696834810893902,
+      "eval_loss": 0.6577129364013672,
+      "eval_mean_token_accuracy": 0.8307611911836332,
+      "eval_num_tokens": 1695212.0,
+      "eval_runtime": 81.951,
+      "eval_samples_per_second": 13.301,
+      "eval_steps_per_second": 1.672,
+      "step": 876
+    },
+    {
+      "entropy": 0.572038494088323,
+      "epoch": 2.0548885077186965,
+      "grad_norm": 0.7493197321891785,
+      "learning_rate": 0.00038234149474685104,
+      "loss": 0.5193290710449219,
+      "mean_token_accuracy": 0.8497188209277119,
+      "num_tokens": 1738782.0,
+      "step": 900
+    },
+    {
+      "entropy": 0.5478819286823273,
+      "epoch": 2.1692395654659804,
+      "grad_norm": 1.2549943923950195,
+      "learning_rate": 0.0003793651001384773,
+      "loss": 0.4793633270263672,
+      "mean_token_accuracy": 0.8575673401355743,
+      "num_tokens": 1837910.0,
+      "step": 950
+    },
+    {
+      "entropy": 0.5431128491461277,
+      "epoch": 2.283590623213265,
+      "grad_norm": 0.8713223338127136,
+      "learning_rate": 0.00037610039945342474,
+      "loss": 0.4849379348754883,
+      "mean_token_accuracy": 0.8562448808550834,
+      "num_tokens": 1937110.0,
+      "step": 1000
+    },
+    {
+      "entropy": 0.5498340383172036,
+      "epoch": 2.3979416809605487,
+      "grad_norm": 0.6592629551887512,
+      "learning_rate": 0.00037255257582546804,
+      "loss": 0.493680419921875,
+      "mean_token_accuracy": 0.854153895676136,
+      "num_tokens": 2036890.0,
+      "step": 1050
+    },
+    {
+      "entropy": 0.5554873237013817,
+      "epoch": 2.512292738707833,
+      "grad_norm": 0.7878748774528503,
+      "learning_rate": 0.0003687272618826019,
+      "loss": 0.4951488494873047,
+      "mean_token_accuracy": 0.8540353265404701,
+      "num_tokens": 2137239.0,
+      "step": 1100
+    },
+    {
+      "entropy": 0.552360344529152,
+      "epoch": 2.6266437964551175,
+      "grad_norm": 0.7380415797233582,
+      "learning_rate": 0.00036463053080451683,
+      "loss": 0.49989810943603513,
+      "mean_token_accuracy": 0.8533406674861908,
+      "num_tokens": 2231081.0,
+      "step": 1150
+    },
+    {
+      "entropy": 0.5345852719247342,
+      "epoch": 2.7409948542024014,
+      "grad_norm": 0.7600613832473755,
+      "learning_rate": 0.0003602688866806424,
+      "loss": 0.5011516189575196,
+      "mean_token_accuracy": 0.8527707865834236,
+      "num_tokens": 2325503.0,
+      "step": 1200
+    },
+    {
+      "entropy": 0.532319468408823,
+      "epoch": 2.8553459119496853,
+      "grad_norm": 0.8197540640830994,
+      "learning_rate": 0.000355649254184066,
+      "loss": 0.5023023605346679,
+      "mean_token_accuracy": 0.8530397671461105,
+      "num_tokens": 2420627.0,
+      "step": 1250
+    },
+    {
+      "entropy": 0.5198138700425625,
+      "epoch": 2.9696969696969697,
+      "grad_norm": 0.8660516142845154,
+      "learning_rate": 0.00035077896757771987,
+      "loss": 0.4941487121582031,
+      "mean_token_accuracy": 0.855004439651966,
+      "num_tokens": 2518051.0,
+      "step": 1300
+    },
+    {
+      "epoch": 3.0,
+      "eval_entropy": 0.5642598084289662,
+      "eval_loss": 0.6125267744064331,
+      "eval_mean_token_accuracy": 0.8375998204641969,
+      "eval_num_tokens": 2542818.0,
+      "eval_runtime": 81.5524,
+      "eval_samples_per_second": 13.366,
+      "eval_steps_per_second": 1.68,
+      "step": 1314
+    },
+    {
+      "entropy": 0.45835615868495805,
+      "epoch": 3.0823327615780447,
+      "grad_norm": 0.734519362449646,
+      "learning_rate": 0.00034566575907029245,
+      "loss": 0.4334117889404297,
+      "mean_token_accuracy": 0.8698135275526095,
+      "num_tokens": 2613861.0,
+      "step": 1350
+    },
+    {
+      "entropy": 0.44648877054452896,
+      "epoch": 3.1966838193253286,
+      "grad_norm": 0.687981128692627,
+      "learning_rate": 0.00034031774654034867,
+      "loss": 0.40152584075927733,
+      "mean_token_accuracy": 0.8759277358651161,
+      "num_tokens": 2709451.0,
+      "step": 1400
+    },
+    {
+      "entropy": 0.4593352344632149,
+      "epoch": 3.311034877072613,
+      "grad_norm": 0.6310222148895264,
+      "learning_rate": 0.0003347434206481497,
+      "loss": 0.41885005950927734,
+      "mean_token_accuracy": 0.8714999267458916,
+      "num_tokens": 2801336.0,
+      "step": 1450
+    },
+    {
+      "entropy": 0.46372457168996334,
+      "epoch": 3.425385934819897,
+      "grad_norm": 0.6795068979263306,
+      "learning_rate": 0.0003289516313556339,
+      "loss": 0.4194014358520508,
+      "mean_token_accuracy": 0.8709167090058326,
+      "num_tokens": 2898945.0,
+      "step": 1500
+    },
+    {
+      "entropy": 0.4548304344713688,
+      "epoch": 3.5397369925671813,
+      "grad_norm": 0.9671820402145386,
+      "learning_rate": 0.00032295157387595997,
+      "loss": 0.4178035736083984,
+      "mean_token_accuracy": 0.8710319435596466,
+      "num_tokens": 2993378.0,
+      "step": 1550
+    },
+    {
+      "entropy": 0.4532127542793751,
+      "epoch": 3.6540880503144653,
+      "grad_norm": 0.7656959295272827,
+      "learning_rate": 0.0003167527740749187,
+      "loss": 0.42567527770996094,
+      "mean_token_accuracy": 0.869033499956131,
+      "num_tokens": 3091323.0,
+      "step": 1600
+    },
+    {
+      "entropy": 0.4460975781828165,
+      "epoch": 3.7684391080617496,
+      "grad_norm": 0.7811372876167297,
+      "learning_rate": 0.0003103650733473916,
+      "loss": 0.4148756790161133,
+      "mean_token_accuracy": 0.8709280017018318,
+      "num_tokens": 3193194.0,
+      "step": 1650
+    },
+    {
+      "entropy": 0.47125090911984446,
+      "epoch": 3.8827901658090336,
+      "grad_norm": 0.8664485216140747,
+      "learning_rate": 0.00030379861299286587,
+      "loss": 0.4209564590454102,
+      "mean_token_accuracy": 0.8695067319273949,
+      "num_tokens": 3292840.0,
+      "step": 1700
+    },
+    {
+      "entropy": 0.4641740974783897,
+      "epoch": 3.997141223556318,
+      "grad_norm": 0.85616534948349,
+      "learning_rate": 0.00029706381811481263,
+      "loss": 0.4319478988647461,
+      "mean_token_accuracy": 0.8689105224609375,
+      "num_tokens": 3387511.0,
+      "step": 1750
+    },
+    {
+      "epoch": 4.0,
+      "eval_entropy": 0.4623802898138979,
+      "eval_loss": 0.6707634925842285,
+      "eval_mean_token_accuracy": 0.8327476525828786,
+      "eval_num_tokens": 3390424.0,
+      "eval_runtime": 81.9106,
+      "eval_samples_per_second": 13.307,
+      "eval_steps_per_second": 1.673,
+      "step": 1752
+    },
+    {
+      "entropy": 0.33750920285125674,
+      "epoch": 4.109777015437393,
+      "grad_norm": 0.6137083172798157,
+      "learning_rate": 0.00029017138106948915,
+      "loss": 0.31130544662475584,
+      "mean_token_accuracy": 0.8997744660692166,
+      "num_tokens": 3483795.0,
+      "step": 1800
+    },
+    {
+      "entropy": 0.3504758469760418,
+      "epoch": 4.224128073184677,
+      "grad_norm": 0.6796069741249084,
+      "learning_rate": 0.00028313224449044337,
+      "loss": 0.31956762313842774,
+      "mean_token_accuracy": 0.895492787361145,
+      "num_tokens": 3576928.0,
+      "step": 1850
+    },
+    {
+      "entropy": 0.34297160781919955,
+      "epoch": 4.338479130931961,
+      "grad_norm": 0.7507141828536987,
+      "learning_rate": 0.00027595758391566995,
+      "loss": 0.31722068786621094,
+      "mean_token_accuracy": 0.8957494455575943,
+      "num_tokens": 3674125.0,
+      "step": 1900
+    },
+    {
+      "entropy": 0.3572557730227709,
+      "epoch": 4.452830188679245,
+      "grad_norm": 0.7509586215019226,
+      "learning_rate": 0.00026865879004500166,
+      "loss": 0.32728019714355466,
+      "mean_token_accuracy": 0.894504280090332,
+      "num_tokens": 3769853.0,
+      "step": 1950
+    },
+    {
+      "entropy": 0.35588217943906786,
+      "epoch": 4.56718124642653,
+      "grad_norm": 0.6168456077575684,
+      "learning_rate": 0.0002612474506559027,
+      "loss": 0.3304343795776367,
+      "mean_token_accuracy": 0.8935999310016632,
+      "num_tokens": 3867574.0,
+      "step": 2000
+    },
+    {
+      "entropy": 0.3554471937566996,
+      "epoch": 4.681532304173814,
+      "grad_norm": 0.6922969222068787,
+      "learning_rate": 0.00025373533220637623,
+      "loss": 0.3315126037597656,
+      "mean_token_accuracy": 0.8925720381736756,
+      "num_tokens": 3969484.0,
+      "step": 2050
+    },
+    {
+      "entropy": 0.35552985966205597,
+      "epoch": 4.795883361921097,
+      "grad_norm": 0.6013802289962769,
+      "learning_rate": 0.00024613436115419326,
+      "loss": 0.3315456771850586,
+      "mean_token_accuracy": 0.8929180511832238,
+      "num_tokens": 4063480.0,
+      "step": 2100
+    },
+    {
+      "entropy": 0.3478537338972092,
+      "epoch": 4.910234419668382,
+      "grad_norm": 0.8090653419494629,
+      "learning_rate": 0.00023845660502210138,
+      "loss": 0.32579368591308594,
+      "mean_token_accuracy": 0.893961855173111,
+      "num_tokens": 4162925.0,
+      "step": 2150
+    },
+    {
+      "epoch": 5.0,
+      "eval_entropy": 0.38475005339532004,
+      "eval_loss": 0.6992377042770386,
+      "eval_mean_token_accuracy": 0.8384414993933518,
+      "eval_num_tokens": 4238030.0,
+      "eval_runtime": 81.1991,
+      "eval_samples_per_second": 13.424,
+      "eval_steps_per_second": 1.687,
+      "step": 2190
+    },
+    {
+      "entropy": 0.3381291570397198,
+      "epoch": 5.022870211549457,
+      "grad_norm": 0.7052969932556152,
+      "learning_rate": 0.00023071425323907478,
+      "loss": 0.3124395370483398,
+      "mean_token_accuracy": 0.8997272504162668,
+      "num_tokens": 4257610.0,
+      "step": 2200
+    },
+    {
+      "entropy": 0.2494013351202011,
+      "epoch": 5.137221269296741,
+      "grad_norm": 0.8653550148010254,
+      "learning_rate": 0.0002229195977880219,
+      "loss": 0.21398515701293946,
+      "mean_token_accuracy": 0.9277590665221215,
+      "num_tokens": 4353703.0,
+      "step": 2250
+    },
+    {
+      "entropy": 0.27507017262279987,
+      "epoch": 5.251572327044025,
+      "grad_norm": 0.7887904644012451,
+      "learning_rate": 0.00021508501369067626,
+      "loss": 0.22058496475219727,
+      "mean_token_accuracy": 0.9253441894054413,
+      "num_tokens": 4449829.0,
+      "step": 2300
+    },
+    {
+      "entropy": 0.2582047886401415,
+      "epoch": 5.3659233847913095,
+      "grad_norm": 0.8728304505348206,
+      "learning_rate": 0.0002072229393606514,
+      "loss": 0.22437936782836915,
+      "mean_token_accuracy": 0.9235965391993523,
+      "num_tokens": 4545644.0,
+      "step": 2350
+    },
+    {
+      "entropy": 0.26232284381985665,
+      "epoch": 5.480274442538594,
+      "grad_norm": 0.6688716411590576,
+      "learning_rate": 0.00019934585685585428,
+      "loss": 0.23260833740234374,
+      "mean_token_accuracy": 0.9217587423324585,
+      "num_tokens": 4637941.0,
+      "step": 2400
+    },
+    {
+      "entropy": 0.2619731902331114,
+      "epoch": 5.594625500285877,
+      "grad_norm": 0.597118616104126,
+      "learning_rate": 0.00019146627206160645,
+      "loss": 0.22075172424316405,
+      "mean_token_accuracy": 0.9252614346146584,
+      "num_tokens": 4738175.0,
+      "step": 2450
+    },
+    {
+      "entropy": 0.26591138120740654,
+      "epoch": 5.708976558033162,
+      "grad_norm": 0.7602102160453796,
+      "learning_rate": 0.00018359669483593703,
+      "loss": 0.22200309753417968,
+      "mean_token_accuracy": 0.9250017151236534,
+      "num_tokens": 4837766.0,
+      "step": 2500
+    },
+    {
+      "entropy": 0.2587306099012494,
+      "epoch": 5.823327615780446,
+      "grad_norm": 0.6875442862510681,
+      "learning_rate": 0.00017574961914856833,
+      "loss": 0.22627996444702148,
+      "mean_token_accuracy": 0.9234379661083222,
+      "num_tokens": 4935670.0,
+      "step": 2550
+    },
+    {
+      "entropy": 0.2689222282171249,
+      "epoch": 5.9376786735277305,
+      "grad_norm": 0.654289186000824,
+      "learning_rate": 0.00016793750324512556,
+      "loss": 0.23186542510986327,
+      "mean_token_accuracy": 0.9214966455101967,
+      "num_tokens": 5030799.0,
+      "step": 2600
+    },
+    {
+      "epoch": 6.0,
+      "eval_entropy": 0.34485617432281046,
+      "eval_loss": 0.7253938317298889,
+      "eval_mean_token_accuracy": 0.8377144193127207,
+      "eval_num_tokens": 5085636.0,
+      "eval_runtime": 81.5281,
+      "eval_samples_per_second": 13.37,
+      "eval_steps_per_second": 1.68,
+      "step": 2628
+    },
+    {
+      "entropy": 0.22255296551334072,
+      "epoch": 6.050314465408805,
+      "grad_norm": 0.5370154976844788,
+      "learning_rate": 0.00016017274986806444,
+      "loss": 0.18560497283935548,
+      "mean_token_accuracy": 0.9371391822238864,
+      "num_tokens": 5126968.0,
+      "step": 2650
+    },
+    {
+      "entropy": 0.1745598777383566,
+      "epoch": 6.164665523156089,
+      "grad_norm": 0.6606088280677795,
+      "learning_rate": 0.00015246768656571565,
+      "loss": 0.14104312896728516,
+      "mean_token_accuracy": 0.9522112870216369,
+      "num_tokens": 5222926.0,
+      "step": 2700
+    },
+    {
+      "entropy": 0.16993012264370919,
+      "epoch": 6.279016580903374,
+      "grad_norm": 0.6963379383087158,
+      "learning_rate": 0.00014483454612071133,
+      "loss": 0.14232961654663087,
+      "mean_token_accuracy": 0.9532496619224549,
+      "num_tokens": 5317743.0,
+      "step": 2750
+    },
+    {
+      "entropy": 0.16726660918444394,
+      "epoch": 6.393367638650657,
+      "grad_norm": 0.7356508374214172,
+      "learning_rate": 0.00013728544712886273,
+      "loss": 0.13951081275939942,
+      "mean_token_accuracy": 0.9533490630984306,
+      "num_tokens": 5415584.0,
+      "step": 2800
+    },
+    {
+      "entropy": 0.17247022703289985,
+      "epoch": 6.507718696397942,
+      "grad_norm": 0.6522324681282043,
+      "learning_rate": 0.00012983237475932584,
+      "loss": 0.13756629943847656,
+      "mean_token_accuracy": 0.9529287284612655,
+      "num_tokens": 5516998.0,
+      "step": 2850
+    },
+    {
+      "entropy": 0.1730361183360219,
+      "epoch": 6.622069754145226,
+      "grad_norm": 0.747295618057251,
+      "learning_rate": 0.00012248716172659724,
+      "loss": 0.1429327392578125,
+      "mean_token_accuracy": 0.9519825592637062,
+      "num_tokens": 5614424.0,
+      "step": 2900
+    },
+    {
+      "entropy": 0.17447776831686496,
+      "epoch": 6.7364208118925095,
+      "grad_norm": 0.624346137046814,
+      "learning_rate": 0.00011526146950455331,
+      "loss": 0.14175209999084473,
+      "mean_token_accuracy": 0.9515316143631936,
+      "num_tokens": 5712047.0,
+      "step": 2950
+    },
+    {
+      "entropy": 0.181520360969007,
+      "epoch": 6.850771869639794,
+      "grad_norm": 0.747399091720581,
+      "learning_rate": 0.00010816676981235426,
+      "loss": 0.14949700355529785,
+      "mean_token_accuracy": 0.9500517123937606,
+      "num_tokens": 5805524.0,
+      "step": 3000
+    },
+    {
+      "entropy": 0.17918075375258924,
+      "epoch": 6.965122927387078,
+      "grad_norm": 0.55885249376297,
+      "learning_rate": 0.00010121432640160985,
+      "loss": 0.14284159660339354,
+      "mean_token_accuracy": 0.9515545222163201,
+      "num_tokens": 5902110.0,
+      "step": 3050
+    },
+    {
+      "epoch": 7.0,
+      "eval_entropy": 0.2829361812476694,
+      "eval_loss": 0.8297234177589417,
+      "eval_mean_token_accuracy": 0.8362262305552072,
+      "eval_num_tokens": 5933242.0,
+      "eval_runtime": 81.6661,
+      "eval_samples_per_second": 13.347,
+      "eval_steps_per_second": 1.678,
+      "step": 3066
+    },
+    {
+      "entropy": 0.13936093423118445,
+      "epoch": 7.077758719268153,
+      "grad_norm": 0.6394845843315125,
+      "learning_rate": 9.441517717371896e-05,
+      "loss": 0.11461298942565917,
+      "mean_token_accuracy": 0.9617519581378414,
+      "num_tokens": 5994998.0,
+      "step": 3100
+    },
+    {
+      "entropy": 0.12445554660633207,
+      "epoch": 7.192109777015437,
+      "grad_norm": 0.3524336814880371,
+      "learning_rate": 8.778011665577682e-05,
+      "loss": 0.09778708457946778,
+      "mean_token_accuracy": 0.9672333124279976,
+      "num_tokens": 6089591.0,
+      "step": 3150
+    },
+    {
+      "entropy": 0.11911977402865886,
+      "epoch": 7.306460834762722,
+      "grad_norm": 0.5336443185806274,
+      "learning_rate": 8.131967886286935e-05,
+      "loss": 0.0957869529724121,
+      "mean_token_accuracy": 0.9681107413768768,
+      "num_tokens": 6189134.0,
+      "step": 3200
+    },
+    {
+      "entropy": 0.12348919529467821,
+      "epoch": 7.420811892510006,
+      "grad_norm": 0.3701528310775757,
+      "learning_rate": 7.504412057396483e-05,
+      "loss": 0.09508245468139648,
+      "mean_token_accuracy": 0.9655187261104584,
+      "num_tokens": 6290034.0,
+      "step": 3250
+    },
+    {
+      "entropy": 0.12596421541646122,
+      "epoch": 7.535162950257289,
+      "grad_norm": 0.4448779225349426,
+      "learning_rate": 6.896340504795279e-05,
+      "loss": 0.09709668159484863,
+      "mean_token_accuracy": 0.9666328871250153,
+      "num_tokens": 6384710.0,
+      "step": 3300
+    },
+    {
+      "entropy": 0.1288480256870389,
+      "epoch": 7.649514008004574,
+      "grad_norm": 0.4604122042655945,
+      "learning_rate": 6.308718620568478e-05,
+      "loss": 0.10019060134887696,
+      "mean_token_accuracy": 0.9648687875270844,
+      "num_tokens": 6480406.0,
+      "step": 3350
+    },
+    {
+      "entropy": 0.11791575787588954,
+      "epoch": 7.763865065751858,
+      "grad_norm": 0.38850095868110657,
+      "learning_rate": 5.742479330312844e-05,
+      "loss": 0.09592323303222657,
+      "mean_token_accuracy": 0.9676644733548164,
+      "num_tokens": 6577776.0,
+      "step": 3400
+    },
+    {
+      "entropy": 0.12499945312738418,
+      "epoch": 7.878216123499143,
+      "grad_norm": 0.2908262014389038,
+      "learning_rate": 5.198521611996973e-05,
+      "loss": 0.09701567649841308,
+      "mean_token_accuracy": 0.9663982436060905,
+      "num_tokens": 6676978.0,
+      "step": 3450
+    },
+    {
+      "entropy": 0.12234330251812935,
+      "epoch": 7.992567181246427,
+      "grad_norm": 0.3459921181201935,
+      "learning_rate": 4.6777090687176785e-05,
+      "loss": 0.0960183334350586,
+      "mean_token_accuracy": 0.967001935839653,
+      "num_tokens": 6775723.0,
+      "step": 3500
+    },
+    {
+      "epoch": 8.0,
+      "eval_entropy": 0.23162500277487902,
+      "eval_loss": 0.9412578344345093,
+      "eval_mean_token_accuracy": 0.8381427713554271,
+      "eval_num_tokens": 6780848.0,
+      "eval_runtime": 81.7803,
+      "eval_samples_per_second": 13.328,
+      "eval_steps_per_second": 1.675,
+      "step": 3504
+    },
+    {
+      "entropy": 0.10137141654787934,
+      "epoch": 8.1052029731275,
+      "grad_norm": 0.26042085886001587,
+      "learning_rate": 4.1808685576186306e-05,
+      "loss": 0.07917721271514892,
+      "mean_token_accuracy": 0.9726437660038169,
+      "num_tokens": 6873669.0,
+      "step": 3550
+    },
+    {
+      "entropy": 0.10200918957591057,
+      "epoch": 8.219554030874786,
+      "grad_norm": 0.24989385902881622,
+      "learning_rate": 3.708788877147874e-05,
+      "loss": 0.07984754085540771,
+      "mean_token_accuracy": 0.9713884252309799,
+      "num_tokens": 6969306.0,
+      "step": 3600
+    },
+    {
+      "entropy": 0.09767990596592427,
+      "epoch": 8.33390508862207,
+      "grad_norm": 0.3814045488834381,
+      "learning_rate": 3.2622195147384906e-05,
+      "loss": 0.07905120849609375,
+      "mean_token_accuracy": 0.9719935733079911,
+      "num_tokens": 7067472.0,
+      "step": 3650
+    },
+    {
+      "entropy": 0.10590592253953218,
+      "epoch": 8.448256146369355,
+      "grad_norm": 0.296535462141037,
+      "learning_rate": 2.8418694569005194e-05,
+      "loss": 0.08534525871276856,
+      "mean_token_accuracy": 0.9699256777763366,
+      "num_tokens": 7160953.0,
+      "step": 3700
+    },
+    {
+      "entropy": 0.10513897916302085,
+      "epoch": 8.562607204116638,
+      "grad_norm": 0.24547256529331207,
+      "learning_rate": 2.448406063613386e-05,
+      "loss": 0.07970548152923584,
+      "mean_token_accuracy": 0.9707170671224594,
+      "num_tokens": 7257979.0,
+      "step": 3750
+    },
+    {
+      "entropy": 0.10294155770912766,
+      "epoch": 8.676958261863922,
+      "grad_norm": 0.195679172873497,
+      "learning_rate": 2.082454008805752e-05,
+      "loss": 0.08084090232849121,
+      "mean_token_accuracy": 0.9708512514829636,
+      "num_tokens": 7355068.0,
+      "step": 3800
+    },
+    {
+      "entropy": 0.10045578340068459,
+      "epoch": 8.791309319611207,
+      "grad_norm": 0.21481232345104218,
+      "learning_rate": 1.7445942886050436e-05,
+      "loss": 0.0806255054473877,
+      "mean_token_accuracy": 0.971126164495945,
+      "num_tokens": 7451068.0,
+      "step": 3850
+    },
+    {
+      "entropy": 0.1046586587280035,
+      "epoch": 8.90566037735849,
+      "grad_norm": 0.34530165791511536,
+      "learning_rate": 1.4353632989310662e-05,
+      "loss": 0.08288439750671386,
+      "mean_token_accuracy": 0.9699718147516251,
+      "num_tokens": 7548327.0,
+      "step": 3900
+    },
+    {
+      "epoch": 9.0,
+      "eval_entropy": 0.2088741992946959,
+      "eval_loss": 1.049438238143921,
+      "eval_mean_token_accuracy": 0.8382186280549878,
+      "eval_num_tokens": 7628454.0,
+      "eval_runtime": 81.7324,
+      "eval_samples_per_second": 13.336,
+      "eval_steps_per_second": 1.676,
+      "step": 3942
+    },
+    {
+      "entropy": 0.09818871334422058,
+      "epoch": 9.018296169239566,
+      "grad_norm": 0.15193390846252441,
+      "learning_rate": 1.1552519838982758e-05,
+      "loss": 0.07954733848571777,
+      "mean_token_accuracy": 0.9727256016682853,
+      "num_tokens": 7645177.0,
+      "step": 3950
+    },
+    {
+      "entropy": 0.0942651672847569,
+      "epoch": 9.13264722698685,
+      "grad_norm": 0.12520743906497955,
+      "learning_rate": 9.047050563785938e-06,
+      "loss": 0.07136336326599121,
+      "mean_token_accuracy": 0.9740171003341674,
+      "num_tokens": 7744145.0,
+      "step": 4000
+    },
+    {
+      "entropy": 0.093073998875916,
+      "epoch": 9.246998284734135,
+      "grad_norm": 0.16596747934818268,
+      "learning_rate": 6.8412029196233095e-06,
+      "loss": 0.07381087303161621,
+      "mean_token_accuracy": 0.9734019112586975,
+      "num_tokens": 7842944.0,
+      "step": 4050
+    },
+    {
+      "entropy": 0.09445923743769527,
+      "epoch": 9.361349342481418,
+      "grad_norm": 0.2622702419757843,
+      "learning_rate": 4.938478974380736e-06,
+      "loss": 0.07485238552093505,
+      "mean_token_accuracy": 0.972542100250721,
+      "num_tokens": 7937715.0,
+      "step": 4100
+    },
+    {
+      "entropy": 0.09793354758992791,
+      "epoch": 9.475700400228702,
+      "grad_norm": 0.20650944113731384,
+      "learning_rate": 3.341899547942234e-06,
+      "loss": 0.0770189094543457,
+      "mean_token_accuracy": 0.9719962891936302,
+      "num_tokens": 8031402.0,
+      "step": 4150
+    },
+    {
+      "entropy": 0.0919106948375702,
+      "epoch": 9.590051457975987,
+      "grad_norm": 0.24635516107082367,
+      "learning_rate": 2.0539994162482062e-06,
+      "loss": 0.073299560546875,
+      "mean_token_accuracy": 0.9736012887954711,
+      "num_tokens": 8127966.0,
+      "step": 4200
+    },
+    {
+      "entropy": 0.0942799466662109,
+      "epoch": 9.70440251572327,
+      "grad_norm": 0.1303575187921524,
+      "learning_rate": 1.0768232870115333e-06,
+      "loss": 0.07512761116027832,
+      "mean_token_accuracy": 0.9723826396465302,
+      "num_tokens": 8223027.0,
+      "step": 4250
+    },
+    {
+      "entropy": 0.0945052900724113,
+      "epoch": 9.818753573470554,
+      "grad_norm": 0.21369652450084686,
+      "learning_rate": 4.1192255348013083e-07,
+      "loss": 0.07480397224426269,
+      "mean_token_accuracy": 0.9724090394377708,
+      "num_tokens": 8318641.0,
+      "step": 4300
+    },
+    {
+      "entropy": 0.08812543807551265,
+      "epoch": 9.933104631217839,
+      "grad_norm": 0.23008032143115997,
+      "learning_rate": 6.035283139993269e-08,
+      "loss": 0.07318362712860108,
+      "mean_token_accuracy": 0.9746180793642998,
+      "num_tokens": 8419263.0,
+      "step": 4350
+    },
+    {
+      "epoch": 10.0,
+      "eval_entropy": 0.20635391673902526,
+      "eval_loss": 1.1006033420562744,
+      "eval_mean_token_accuracy": 0.8381778462959902,
+      "eval_num_tokens": 8476060.0,
+      "eval_runtime": 81.4863,
+      "eval_samples_per_second": 13.376,
+      "eval_steps_per_second": 1.681,
+      "step": 4380
+    }
+  ],
+  "logging_steps": 50,
+  "max_steps": 4380,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 10,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 3.5225312017032806e+17,
+  "train_batch_size": 4,
+  "trial_name": null,
+  "trial_params": null
+}

systematicity_original_Swedish/Qwen3.5-4B-Base_systematicity_splits_original_features_train_systematicity_splits_original_features_test1/checkpoint-876/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

systematicity_original_Swedish/Qwen3.5-4B-Base_systematicity_splits_original_features_train_systematicity_splits_original_features_test1/checkpoint-876/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": 0.026477173062682283,
+  "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": [
+    "q_proj",
+    "k_proj",
+    "down_proj",
+    "v_proj",
+    "gate_proj",
+    "o_proj",
+    "up_proj"
+  ],
+  "target_parameters": null,
+  "task_type": "CAUSAL_LM",
+  "trainable_token_indices": null,
+  "use_dora": false,
+  "use_qalora": false,
+  "use_rslora": false
+}

systematicity_original_Swedish/Qwen3.5-4B-Base_systematicity_splits_original_features_train_systematicity_splits_original_features_test1/checkpoint-876/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 %}

systematicity_original_Swedish/Qwen3.5-4B-Base_systematicity_splits_original_features_train_systematicity_splits_original_features_test2/checkpoint-4380/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 %}

systematicity_original_Swedish/Qwen3.5-4B-Base_systematicity_splits_original_features_train_systematicity_splits_original_features_test2/checkpoint-4380/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|>"
+}

systematicity_original_Swedish/Qwen3.5-4B-Base_systematicity_splits_original_features_train_systematicity_splits_original_features_test2/checkpoint-4380/trainer_state.json

@@ -0,0 +1,1014 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 10.0,
+  "eval_steps": 500,
+  "global_step": 4380,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "entropy": 2.00145958840847,
+      "epoch": 0.11435105774728416,
+      "grad_norm": 2.497065782546997,
+      "learning_rate": 4.571087888923681e-06,
+      "loss": 1.7145748901367188,
+      "mean_token_accuracy": 0.6452348600327968,
+      "num_tokens": 96586.0,
+      "step": 50
+    },
+    {
+      "entropy": 0.8970040166378022,
+      "epoch": 0.22870211549456831,
+      "grad_norm": 1.9262803792953491,
+      "learning_rate": 9.23546328578458e-06,
+      "loss": 0.7682270050048828,
+      "mean_token_accuracy": 0.7993282404541969,
+      "num_tokens": 190720.0,
+      "step": 100
+    },
+    {
+      "entropy": 0.7573070672154426,
+      "epoch": 0.34305317324185247,
+      "grad_norm": 1.519248366355896,
+      "learning_rate": 1.389983868264548e-05,
+      "loss": 0.6529589080810547,
+      "mean_token_accuracy": 0.8195512276887894,
+      "num_tokens": 288515.0,
+      "step": 150
+    },
+    {
+      "entropy": 0.7158778738975525,
+      "epoch": 0.45740423098913663,
+      "grad_norm": 1.4797775745391846,
+      "learning_rate": 1.8564214079506377e-05,
+      "loss": 0.6193486022949218,
+      "mean_token_accuracy": 0.827950005531311,
+      "num_tokens": 387874.0,
+      "step": 200
+    },
+    {
+      "entropy": 0.6906554877758027,
+      "epoch": 0.5717552887364208,
+      "grad_norm": 1.5661845207214355,
+      "learning_rate": 2.322858947636728e-05,
+      "loss": 0.5904607009887696,
+      "mean_token_accuracy": 0.8340386778116227,
+      "num_tokens": 487865.0,
+      "step": 250
+    },
+    {
+      "entropy": 0.6806062492728233,
+      "epoch": 0.6861063464837049,
+      "grad_norm": 1.276515245437622,
+      "learning_rate": 2.7892964873228177e-05,
+      "loss": 0.5840428924560547,
+      "mean_token_accuracy": 0.8354081481695175,
+      "num_tokens": 584269.0,
+      "step": 300
+    },
+    {
+      "entropy": 0.6750492942333222,
+      "epoch": 0.8004574042309891,
+      "grad_norm": 1.3048511743545532,
+      "learning_rate": 3.2557340270089076e-05,
+      "loss": 0.5731664657592773,
+      "mean_token_accuracy": 0.8381186512112617,
+      "num_tokens": 676067.0,
+      "step": 350
+    },
+    {
+      "entropy": 0.652309263497591,
+      "epoch": 0.9148084619782733,
+      "grad_norm": 1.1547592878341675,
+      "learning_rate": 3.722171566694997e-05,
+      "loss": 0.5572821807861328,
+      "mean_token_accuracy": 0.8416853401064873,
+      "num_tokens": 774009.0,
+      "step": 400
+    },
+    {
+      "epoch": 1.0,
+      "eval_entropy": 0.5965869100424495,
+      "eval_loss": 0.5746821761131287,
+      "eval_mean_token_accuracy": 0.8380154971658749,
+      "eval_num_tokens": 847606.0,
+      "eval_runtime": 75.0294,
+      "eval_samples_per_second": 14.514,
+      "eval_steps_per_second": 1.826,
+      "step": 438
+    },
+    {
+      "entropy": 0.6367223400452415,
+      "epoch": 1.0274442538593482,
+      "grad_norm": 0.936708390712738,
+      "learning_rate": 4.085914344568524e-05,
+      "loss": 0.5404758834838868,
+      "mean_token_accuracy": 0.846109490104133,
+      "num_tokens": 870214.0,
+      "step": 450
+    },
+    {
+      "entropy": 0.5927853919565678,
+      "epoch": 1.1417953116066324,
+      "grad_norm": 1.0234040021896362,
+      "learning_rate": 4.0835791756497073e-05,
+      "loss": 0.5002776718139649,
+      "mean_token_accuracy": 0.8537301248311997,
+      "num_tokens": 971584.0,
+      "step": 500
+    },
+    {
+      "entropy": 0.5995605960488319,
+      "epoch": 1.2561463693539165,
+      "grad_norm": 0.9836112856864929,
+      "learning_rate": 4.0780043181954356e-05,
+      "loss": 0.5042121124267578,
+      "mean_token_accuracy": 0.8527141407132148,
+      "num_tokens": 1063906.0,
+      "step": 550
+    },
+    {
+      "entropy": 0.5905857393145562,
+      "epoch": 1.3704974271012007,
+      "grad_norm": 1.0268890857696533,
+      "learning_rate": 4.0691986230115715e-05,
+      "loss": 0.5004239654541016,
+      "mean_token_accuracy": 0.8539833968877792,
+      "num_tokens": 1159282.0,
+      "step": 600
+    },
+    {
+      "entropy": 0.6022135411202908,
+      "epoch": 1.4848484848484849,
+      "grad_norm": 0.9739264845848083,
+      "learning_rate": 4.0571760702759385e-05,
+      "loss": 0.4968523788452148,
+      "mean_token_accuracy": 0.854298751950264,
+      "num_tokens": 1254998.0,
+      "step": 650
+    },
+    {
+      "entropy": 0.5831877724826335,
+      "epoch": 1.599199542595769,
+      "grad_norm": 0.8632996082305908,
+      "learning_rate": 4.0419557473429846e-05,
+      "loss": 0.49389530181884767,
+      "mean_token_accuracy": 0.8568908262252808,
+      "num_tokens": 1351877.0,
+      "step": 700
+    },
+    {
+      "entropy": 0.5858544714748859,
+      "epoch": 1.7135506003430532,
+      "grad_norm": 0.8835417628288269,
+      "learning_rate": 4.023561818440143e-05,
+      "loss": 0.49105472564697267,
+      "mean_token_accuracy": 0.8568279257416725,
+      "num_tokens": 1448801.0,
+      "step": 750
+    },
+    {
+      "entropy": 0.587700717896223,
+      "epoch": 1.8279016580903373,
+      "grad_norm": 0.8575327396392822,
+      "learning_rate": 4.002023486304002e-05,
+      "loss": 0.4949332046508789,
+      "mean_token_accuracy": 0.8551343831419945,
+      "num_tokens": 1547040.0,
+      "step": 800
+    },
+    {
+      "entropy": 0.5782938550412655,
+      "epoch": 1.9422527158376215,
+      "grad_norm": 0.7499766945838928,
+      "learning_rate": 3.9773749458171914e-05,
+      "loss": 0.4910758972167969,
+      "mean_token_accuracy": 0.8579927015304566,
+      "num_tokens": 1646444.0,
+      "step": 850
+    },
+    {
+      "epoch": 2.0,
+      "eval_entropy": 0.5206899273134497,
+      "eval_loss": 0.5700324177742004,
+      "eval_mean_token_accuracy": 0.8452296922676754,
+      "eval_num_tokens": 1695212.0,
+      "eval_runtime": 74.7024,
+      "eval_samples_per_second": 14.578,
+      "eval_steps_per_second": 1.834,
+      "step": 876
+    },
+    {
+      "entropy": 0.5343730554060282,
+      "epoch": 2.0548885077186965,
+      "grad_norm": 0.9055376648902893,
+      "learning_rate": 3.949655329719591e-05,
+      "loss": 0.4389276885986328,
+      "mean_token_accuracy": 0.8690450318573695,
+      "num_tokens": 1738782.0,
+      "step": 900
+    },
+    {
+      "entropy": 0.5053752763569355,
+      "epoch": 2.1692395654659804,
+      "grad_norm": 0.8962181806564331,
+      "learning_rate": 3.918908646480056e-05,
+      "loss": 0.40234622955322263,
+      "mean_token_accuracy": 0.8774205523729325,
+      "num_tokens": 1837910.0,
+      "step": 950
+    },
+    {
+      "entropy": 0.5023033122718334,
+      "epoch": 2.283590623213265,
+      "grad_norm": 1.0863624811172485,
+      "learning_rate": 3.885183710427288e-05,
+      "loss": 0.40644466400146484,
+      "mean_token_accuracy": 0.8757456785440445,
+      "num_tokens": 1937110.0,
+      "step": 1000
+    },
+    {
+      "entropy": 0.5023509474098682,
+      "epoch": 2.3979416809605487,
+      "grad_norm": 0.7502025961875916,
+      "learning_rate": 3.8485340642507936e-05,
+      "loss": 0.4057846450805664,
+      "mean_token_accuracy": 0.8757887029647827,
+      "num_tokens": 2036890.0,
+      "step": 1050
+    },
+    {
+      "entropy": 0.49996689677238465,
+      "epoch": 2.512292738707833,
+      "grad_norm": 0.925940752029419,
+      "learning_rate": 3.8090178939949456e-05,
+      "loss": 0.4050485610961914,
+      "mean_token_accuracy": 0.8761042952537537,
+      "num_tokens": 2137239.0,
+      "step": 1100
+    },
+    {
+      "entropy": 0.507868722230196,
+      "epoch": 2.6266437964551175,
+      "grad_norm": 1.1511523723602295,
+      "learning_rate": 3.766697936681131e-05,
+      "loss": 0.40807209014892576,
+      "mean_token_accuracy": 0.875197849869728,
+      "num_tokens": 2231081.0,
+      "step": 1150
+    },
+    {
+      "entropy": 0.5105524662137032,
+      "epoch": 2.7409948542024014,
+      "grad_norm": 1.1561952829360962,
+      "learning_rate": 3.72164138070463e-05,
+      "loss": 0.40918811798095706,
+      "mean_token_accuracy": 0.8738781309127808,
+      "num_tokens": 2325503.0,
+      "step": 1200
+    },
+    {
+      "entropy": 0.5113652451336383,
+      "epoch": 2.8553459119496853,
+      "grad_norm": 1.0278860330581665,
+      "learning_rate": 3.673919759164364e-05,
+      "loss": 0.4089116287231445,
+      "mean_token_accuracy": 0.8750931170582771,
+      "num_tokens": 2420627.0,
+      "step": 1250
+    },
+    {
+      "entropy": 0.49678863406181334,
+      "epoch": 2.9696969696969697,
+      "grad_norm": 0.970623254776001,
+      "learning_rate": 3.623608836294867e-05,
+      "loss": 0.40086280822753906,
+      "mean_token_accuracy": 0.8785448360443116,
+      "num_tokens": 2518051.0,
+      "step": 1300
+    },
+    {
+      "epoch": 3.0,
+      "eval_entropy": 0.4818899542310812,
+      "eval_loss": 0.5618253350257874,
+      "eval_mean_token_accuracy": 0.8463477844739482,
+      "eval_num_tokens": 2542818.0,
+      "eval_runtime": 74.5733,
+      "eval_samples_per_second": 14.603,
+      "eval_steps_per_second": 1.837,
+      "step": 1314
+    },
+    {
+      "entropy": 0.44624785980597365,
+      "epoch": 3.0823327615780447,
+      "grad_norm": 1.0951991081237793,
+      "learning_rate": 3.570788487180786e-05,
+      "loss": 0.34254161834716795,
+      "mean_token_accuracy": 0.8942674825034166,
+      "num_tokens": 2613861.0,
+      "step": 1350
+    },
+    {
+      "entropy": 0.41790418803691864,
+      "epoch": 3.1966838193253286,
+      "grad_norm": 1.0191915035247803,
+      "learning_rate": 3.51554257094487e-05,
+      "loss": 0.31106861114501955,
+      "mean_token_accuracy": 0.902518424987793,
+      "num_tokens": 2709451.0,
+      "step": 1400
+    },
+    {
+      "entropy": 0.42402541801333427,
+      "epoch": 3.311034877072613,
+      "grad_norm": 0.9670789241790771,
+      "learning_rate": 3.4579587976107854e-05,
+      "loss": 0.31938417434692384,
+      "mean_token_accuracy": 0.8987989267706871,
+      "num_tokens": 2801336.0,
+      "step": 1450
+    },
+    {
+      "entropy": 0.4250092693418264,
+      "epoch": 3.425385934819897,
+      "grad_norm": 0.9686717987060547,
+      "learning_rate": 3.398128588852137e-05,
+      "loss": 0.3194206428527832,
+      "mean_token_accuracy": 0.8993277350068092,
+      "num_tokens": 2898945.0,
+      "step": 1500
+    },
+    {
+      "entropy": 0.42681451261043546,
+      "epoch": 3.5397369925671813,
+      "grad_norm": 1.1068836450576782,
+      "learning_rate": 3.3361469328487553e-05,
+      "loss": 0.3177553367614746,
+      "mean_token_accuracy": 0.8984623408317566,
+      "num_tokens": 2993378.0,
+      "step": 1550
+    },
+    {
+      "entropy": 0.4175560978055,
+      "epoch": 3.6540880503144653,
+      "grad_norm": 1.0208574533462524,
+      "learning_rate": 3.2721122334807005e-05,
+      "loss": 0.3182306098937988,
+      "mean_token_accuracy": 0.8999121606349945,
+      "num_tokens": 3091323.0,
+      "step": 1600
+    },
+    {
+      "entropy": 0.41283833995461466,
+      "epoch": 3.7684391080617496,
+      "grad_norm": 1.1925355195999146,
+      "learning_rate": 3.206126154099399e-05,
+      "loss": 0.31240909576416015,
+      "mean_token_accuracy": 0.9010123562812805,
+      "num_tokens": 3193194.0,
+      "step": 1650
+    },
+    {
+      "entropy": 0.41667531803250313,
+      "epoch": 3.8827901658090336,
+      "grad_norm": 1.2467414140701294,
+      "learning_rate": 3.13829345612395e-05,
+      "loss": 0.31813507080078124,
+      "mean_token_accuracy": 0.8984718990325927,
+      "num_tokens": 3292840.0,
+      "step": 1700
+    },
+    {
+      "entropy": 0.43274998769164086,
+      "epoch": 3.997141223556318,
+      "grad_norm": 1.203948974609375,
+      "learning_rate": 3.068721832718849e-05,
+      "loss": 0.3260676956176758,
+      "mean_token_accuracy": 0.8978011286258698,
+      "num_tokens": 3387511.0,
+      "step": 1750
+    },
+    {
+      "epoch": 4.0,
+      "eval_entropy": 0.41724466037576213,
+      "eval_loss": 0.6267160773277283,
+      "eval_mean_token_accuracy": 0.8441739900268778,
+      "eval_num_tokens": 3390424.0,
+      "eval_runtime": 74.4855,
+      "eval_samples_per_second": 14.62,
+      "eval_steps_per_second": 1.839,
+      "step": 1752
+    },
+    {
+      "entropy": 0.34096736373937675,
+      "epoch": 4.109777015437393,
+      "grad_norm": 1.1315901279449463,
+      "learning_rate": 2.9975217378171883e-05,
+      "loss": 0.22814653396606446,
+      "mean_token_accuracy": 0.927253995450015,
+      "num_tokens": 3483795.0,
+      "step": 1800
+    },
+    {
+      "entropy": 0.3438922663778067,
+      "epoch": 4.224128073184677,
+      "grad_norm": 1.2761495113372803,
+      "learning_rate": 2.92480621076078e-05,
+      "loss": 0.22710008621215821,
+      "mean_token_accuracy": 0.9249320811033249,
+      "num_tokens": 3576928.0,
+      "step": 1850
+    },
+    {
+      "entropy": 0.33149779081344605,
+      "epoch": 4.338479130931961,
+      "grad_norm": 1.3810840845108032,
+      "learning_rate": 2.8506906968356037e-05,
+      "loss": 0.22302957534790038,
+      "mean_token_accuracy": 0.9278453385829926,
+      "num_tokens": 3674125.0,
+      "step": 1900
+    },
+    {
+      "entropy": 0.33706638153642415,
+      "epoch": 4.452830188679245,
+      "grad_norm": 1.2733014822006226,
+      "learning_rate": 2.7752928639875193e-05,
+      "loss": 0.22845399856567383,
+      "mean_token_accuracy": 0.9249477565288544,
+      "num_tokens": 3769853.0,
+      "step": 1950
+    },
+    {
+      "entropy": 0.33741924159228803,
+      "epoch": 4.56718124642653,
+      "grad_norm": 1.5695229768753052,
+      "learning_rate": 2.69873241600921e-05,
+      "loss": 0.22755613327026367,
+      "mean_token_accuracy": 0.9242416375875473,
+      "num_tokens": 3867574.0,
+      "step": 2000
+    },
+    {
+      "entropy": 0.3302827900648117,
+      "epoch": 4.681532304173814,
+      "grad_norm": 1.6182698011398315,
+      "learning_rate": 2.621130902494958e-05,
+      "loss": 0.2259531021118164,
+      "mean_token_accuracy": 0.9267010247707367,
+      "num_tokens": 3969484.0,
+      "step": 2050
+    },
+    {
+      "entropy": 0.33573394276201723,
+      "epoch": 4.795883361921097,
+      "grad_norm": 1.0237946510314941,
+      "learning_rate": 2.5426115258649745e-05,
+      "loss": 0.23189125061035157,
+      "mean_token_accuracy": 0.9252221485972405,
+      "num_tokens": 4063480.0,
+      "step": 2100
+    },
+    {
+      "entropy": 0.3275442124158144,
+      "epoch": 4.910234419668382,
+      "grad_norm": 1.2637572288513184,
+      "learning_rate": 2.463298945765653e-05,
+      "loss": 0.22184085845947266,
+      "mean_token_accuracy": 0.926847530901432,
+      "num_tokens": 4162925.0,
+      "step": 2150
+    },
+    {
+      "epoch": 5.0,
+      "eval_entropy": 0.3566678982146465,
+      "eval_loss": 0.7187316417694092,
+      "eval_mean_token_accuracy": 0.8409678413920159,
+      "eval_num_tokens": 4238030.0,
+      "eval_runtime": 74.3527,
+      "eval_samples_per_second": 14.646,
+      "eval_steps_per_second": 1.843,
+      "step": 2190
+    },
+    {
+      "entropy": 0.32694301357124056,
+      "epoch": 5.022870211549457,
+      "grad_norm": 1.3936444520950317,
+      "learning_rate": 2.383319081156289e-05,
+      "loss": 0.21443597793579103,
+      "mean_token_accuracy": 0.9309739987862292,
+      "num_tokens": 4257610.0,
+      "step": 2200
+    },
+    {
+      "entropy": 0.25806200847029687,
+      "epoch": 5.137221269296741,
+      "grad_norm": 1.5030523538589478,
+      "learning_rate": 2.3027989103964757e-05,
+      "loss": 0.14753504753112792,
+      "mean_token_accuracy": 0.9510083213448525,
+      "num_tokens": 4353703.0,
+      "step": 2250
+    },
+    {
+      "entropy": 0.25820051588118076,
+      "epoch": 5.251572327044025,
+      "grad_norm": 1.6646113395690918,
+      "learning_rate": 2.2218662696515684e-05,
+      "loss": 0.14954824447631837,
+      "mean_token_accuracy": 0.9509174367785453,
+      "num_tokens": 4449829.0,
+      "step": 2300
+    },
+    {
+      "entropy": 0.26250154070556164,
+      "epoch": 5.3659233847913095,
+      "grad_norm": 1.2992604970932007,
+      "learning_rate": 2.140649649936269e-05,
+      "loss": 0.15098441123962403,
+      "mean_token_accuracy": 0.9509531611204147,
+      "num_tokens": 4545644.0,
+      "step": 2350
+    },
+    {
+      "entropy": 0.2741938025504351,
+      "epoch": 5.480274442538594,
+      "grad_norm": 1.5651607513427734,
+      "learning_rate": 2.059277993118554e-05,
+      "loss": 0.15704907417297365,
+      "mean_token_accuracy": 0.9487551787495613,
+      "num_tokens": 4637941.0,
+      "step": 2400
+    },
+    {
+      "entropy": 0.2517349097877741,
+      "epoch": 5.594625500285877,
+      "grad_norm": 1.2501096725463867,
+      "learning_rate": 1.9778804872078128e-05,
+      "loss": 0.14757199287414552,
+      "mean_token_accuracy": 0.9515560516715049,
+      "num_tokens": 4738175.0,
+      "step": 2450
+    },
+    {
+      "entropy": 0.24905245147645474,
+      "epoch": 5.708976558033162,
+      "grad_norm": 1.4037293195724487,
+      "learning_rate": 1.896586361252207e-05,
+      "loss": 0.1481949520111084,
+      "mean_token_accuracy": 0.9518148481845856,
+      "num_tokens": 4837766.0,
+      "step": 2500
+    },
+    {
+      "entropy": 0.2581802429258823,
+      "epoch": 5.823327615780446,
+      "grad_norm": 1.4516733884811401,
+      "learning_rate": 1.8155246801708754e-05,
+      "loss": 0.15240735054016114,
+      "mean_token_accuracy": 0.9496219110488892,
+      "num_tokens": 4935670.0,
+      "step": 2550
+    },
+    {
+      "entropy": 0.26441981591284275,
+      "epoch": 5.9376786735277305,
+      "grad_norm": 1.4071907997131348,
+      "learning_rate": 1.7348241398467104e-05,
+      "loss": 0.1547057342529297,
+      "mean_token_accuracy": 0.9486072099208832,
+      "num_tokens": 5030799.0,
+      "step": 2600
+    },
+    {
+      "epoch": 6.0,
+      "eval_entropy": 0.3113773436876979,
+      "eval_loss": 0.7981818318367004,
+      "eval_mean_token_accuracy": 0.8384634600068531,
+      "eval_num_tokens": 5085636.0,
+      "eval_runtime": 74.6627,
+      "eval_samples_per_second": 14.586,
+      "eval_steps_per_second": 1.835,
+      "step": 2628
+    },
+    {
+      "entropy": 0.23431323147213398,
+      "epoch": 6.050314465408805,
+      "grad_norm": 1.4807639122009277,
+      "learning_rate": 1.6546128628050357e-05,
+      "loss": 0.12903169631958009,
+      "mean_token_accuracy": 0.9579363782998874,
+      "num_tokens": 5126968.0,
+      "step": 2650
+    },
+    {
+      "entropy": 0.21113360580056906,
+      "epoch": 6.164665523156089,
+      "grad_norm": 1.0075150728225708,
+      "learning_rate": 1.5750181948025524e-05,
+      "loss": 0.10537015914916992,
+      "mean_token_accuracy": 0.9646552309393883,
+      "num_tokens": 5222926.0,
+      "step": 2700
+    },
+    {
+      "entropy": 0.2040509721636772,
+      "epoch": 6.279016580903374,
+      "grad_norm": 1.2175794839859009,
+      "learning_rate": 1.4961665026495188e-05,
+      "loss": 0.107632474899292,
+      "mean_token_accuracy": 0.9661425757408142,
+      "num_tokens": 5317743.0,
+      "step": 2750
+    },
+    {
+      "entropy": 0.20550984337925912,
+      "epoch": 6.393367638650657,
+      "grad_norm": 1.7824593782424927,
+      "learning_rate": 1.4181829735861169e-05,
+      "loss": 0.10691510200500488,
+      "mean_token_accuracy": 0.9654685056209564,
+      "num_tokens": 5415584.0,
+      "step": 2800
+    },
+    {
+      "entropy": 0.20776449400931596,
+      "epoch": 6.507718696397942,
+      "grad_norm": 1.4376789331436157,
+      "learning_rate": 1.3411914165315586e-05,
+      "loss": 0.10438115119934083,
+      "mean_token_accuracy": 0.965470227599144,
+      "num_tokens": 5516998.0,
+      "step": 2850
+    },
+    {
+      "entropy": 0.20203392535448075,
+      "epoch": 6.622069754145226,
+      "grad_norm": 1.2457393407821655,
+      "learning_rate": 1.2653140655214344e-05,
+      "loss": 0.10661219596862793,
+      "mean_token_accuracy": 0.9654071775078773,
+      "num_tokens": 5614424.0,
+      "step": 2900
+    },
+    {
+      "entropy": 0.20881931744515897,
+      "epoch": 6.7364208118925095,
+      "grad_norm": 1.3488210439682007,
+      "learning_rate": 1.190671385645412e-05,
+      "loss": 0.10760212898254394,
+      "mean_token_accuracy": 0.9644413706660271,
+      "num_tokens": 5712047.0,
+      "step": 2950
+    },
+    {
+      "entropy": 0.22206872243434192,
+      "epoch": 6.850771869639794,
+      "grad_norm": 1.450213074684143,
+      "learning_rate": 1.1173818817933463e-05,
+      "loss": 0.11339238166809082,
+      "mean_token_accuracy": 0.9621626555919647,
+      "num_tokens": 5805524.0,
+      "step": 3000
+    },
+    {
+      "entropy": 0.21174554128199816,
+      "epoch": 6.965122927387078,
+      "grad_norm": 1.0198427438735962,
+      "learning_rate": 1.0455619105134785e-05,
+      "loss": 0.10970592498779297,
+      "mean_token_accuracy": 0.9635774600505829,
+      "num_tokens": 5902110.0,
+      "step": 3050
+    },
+    {
+      "epoch": 7.0,
+      "eval_entropy": 0.2774139005337318,
+      "eval_loss": 0.915586531162262,
+      "eval_mean_token_accuracy": 0.8365351622992189,
+      "eval_num_tokens": 5933242.0,
+      "eval_runtime": 74.6474,
+      "eval_samples_per_second": 14.589,
+      "eval_steps_per_second": 1.835,
+      "step": 3066
+    },
+    {
+      "entropy": 0.1982341786931614,
+      "epoch": 7.077758719268153,
+      "grad_norm": 0.7468180656433105,
+      "learning_rate": 9.753254952813877e-06,
+      "loss": 0.09601005554199218,
+      "mean_token_accuracy": 0.9682205177200627,
+      "num_tokens": 5994998.0,
+      "step": 3100
+    },
+    {
+      "entropy": 0.19256796240806578,
+      "epoch": 7.192109777015437,
+      "grad_norm": 0.8898351788520813,
+      "learning_rate": 9.067841454730092e-06,
+      "loss": 0.08880398750305175,
+      "mean_token_accuracy": 0.9702904996275902,
+      "num_tokens": 6089591.0,
+      "step": 3150
+    },
+    {
+      "entropy": 0.17841659400612117,
+      "epoch": 7.306460834762722,
+      "grad_norm": 1.2330012321472168,
+      "learning_rate": 8.40046679329102e-06,
+      "loss": 0.08539151191711426,
+      "mean_token_accuracy": 0.9713234108686447,
+      "num_tokens": 6189134.0,
+      "step": 3200
+    },
+    {
+      "entropy": 0.19028046015650035,
+      "epoch": 7.420811892510006,
+      "grad_norm": 0.7215322256088257,
+      "learning_rate": 7.752190511922482e-06,
+      "loss": 0.08564237594604492,
+      "mean_token_accuracy": 0.9689799511432647,
+      "num_tokens": 6290034.0,
+      "step": 3250
+    },
+    {
+      "entropy": 0.1919882706925273,
+      "epoch": 7.535162950257289,
+      "grad_norm": 1.3013434410095215,
+      "learning_rate": 7.124041832906524e-06,
+      "loss": 0.08905813217163086,
+      "mean_token_accuracy": 0.9699208292365075,
+      "num_tokens": 6384710.0,
+      "step": 3300
+    },
+    {
+      "entropy": 0.1951625467836857,
+      "epoch": 7.649514008004574,
+      "grad_norm": 1.1525713205337524,
+      "learning_rate": 6.517018023358224e-06,
+      "loss": 0.09026041984558106,
+      "mean_token_accuracy": 0.9685596722364426,
+      "num_tokens": 6480406.0,
+      "step": 3350
+    },
+    {
+      "entropy": 0.184615648239851,
+      "epoch": 7.763865065751858,
+      "grad_norm": 0.6319608688354492,
+      "learning_rate": 5.9320828119353666e-06,
+      "loss": 0.08877361297607422,
+      "mean_token_accuracy": 0.970262805223465,
+      "num_tokens": 6577776.0,
+      "step": 3400
+    },
+    {
+      "entropy": 0.18618743564933538,
+      "epoch": 7.878216123499143,
+      "grad_norm": 0.6197832226753235,
+      "learning_rate": 5.370164858794843e-06,
+      "loss": 0.08898484230041503,
+      "mean_token_accuracy": 0.9694998559355735,
+      "num_tokens": 6676978.0,
+      "step": 3450
+    },
+    {
+      "entropy": 0.18308176279067992,
+      "epoch": 7.992567181246427,
+      "grad_norm": 0.7581543326377869,
+      "learning_rate": 4.83215628122472e-06,
+      "loss": 0.08884740829467773,
+      "mean_token_accuracy": 0.9700374433398247,
+      "num_tokens": 6775723.0,
+      "step": 3500
+    },
+    {
+      "epoch": 8.0,
+      "eval_entropy": 0.24912942369488905,
+      "eval_loss": 1.0281336307525635,
+      "eval_mean_token_accuracy": 0.8355324799126952,
+      "eval_num_tokens": 6780848.0,
+      "eval_runtime": 74.4179,
+      "eval_samples_per_second": 14.634,
+      "eval_steps_per_second": 1.841,
+      "step": 3504
+    },
+    {
+      "entropy": 0.17291418394915342,
+      "epoch": 8.1052029731275,
+      "grad_norm": 0.528134286403656,
+      "learning_rate": 4.318911237292925e-06,
+      "loss": 0.07714019298553466,
+      "mean_token_accuracy": 0.9734989068834915,
+      "num_tokens": 6873669.0,
+      "step": 3550
+    },
+    {
+      "entropy": 0.18380620382726193,
+      "epoch": 8.219554030874786,
+      "grad_norm": 0.6781896352767944,
+      "learning_rate": 3.831244569761018e-06,
+      "loss": 0.07927018165588379,
+      "mean_token_accuracy": 0.9721345752477646,
+      "num_tokens": 6969306.0,
+      "step": 3600
+    },
+    {
+      "entropy": 0.1734279589354992,
+      "epoch": 8.33390508862207,
+      "grad_norm": 0.6227915287017822,
+      "learning_rate": 3.3699305124161535e-06,
+      "loss": 0.07849381923675537,
+      "mean_token_accuracy": 0.9724716311693191,
+      "num_tokens": 7067472.0,
+      "step": 3650
+    },
+    {
+      "entropy": 0.18374973967671393,
+      "epoch": 8.448256146369355,
+      "grad_norm": 0.5389400124549866,
+      "learning_rate": 2.935701460874958e-06,
+      "loss": 0.08503846168518066,
+      "mean_token_accuracy": 0.970702798962593,
+      "num_tokens": 7160953.0,
+      "step": 3700
+    },
+    {
+      "entropy": 0.18278369564563035,
+      "epoch": 8.562607204116638,
+      "grad_norm": 0.5344480276107788,
+      "learning_rate": 2.5292468098109876e-06,
+      "loss": 0.07878175735473633,
+      "mean_token_accuracy": 0.9714985090494156,
+      "num_tokens": 7257979.0,
+      "step": 3750
+    },
+    {
+      "entropy": 0.1782318002358079,
+      "epoch": 8.676958261863922,
+      "grad_norm": 1.3235697746276855,
+      "learning_rate": 2.151211858451654e-06,
+      "loss": 0.07996997356414795,
+      "mean_token_accuracy": 0.9716796427965164,
+      "num_tokens": 7355068.0,
+      "step": 3800
+    },
+    {
+      "entropy": 0.17475281044840812,
+      "epoch": 8.791309319611207,
+      "grad_norm": 0.6435418128967285,
+      "learning_rate": 1.802196786082429e-06,
+      "loss": 0.08047654151916504,
+      "mean_token_accuracy": 0.9712266248464584,
+      "num_tokens": 7451068.0,
+      "step": 3850
+    },
+    {
+      "entropy": 0.17993240412324668,
+      "epoch": 8.90566037735849,
+      "grad_norm": 0.6569116115570068,
+      "learning_rate": 1.48275569918472e-06,
+      "loss": 0.08242264747619629,
+      "mean_token_accuracy": 0.9710843190550804,
+      "num_tokens": 7548327.0,
+      "step": 3900
+    },
+    {
+      "epoch": 9.0,
+      "eval_entropy": 0.2359609913869496,
+      "eval_loss": 1.1114780902862549,
+      "eval_mean_token_accuracy": 0.8357110419412599,
+      "eval_num_tokens": 7628454.0,
+      "eval_runtime": 74.4085,
+      "eval_samples_per_second": 14.635,
+      "eval_steps_per_second": 1.841,
+      "step": 3942
+    },
+    {
+      "entropy": 0.1732393957409762,
+      "epoch": 9.018296169239566,
+      "grad_norm": 0.3653605282306671,
+      "learning_rate": 1.1933957517203372e-06,
+      "loss": 0.07902424335479737,
+      "mean_token_accuracy": 0.9731994507276467,
+      "num_tokens": 7645177.0,
+      "step": 3950
+    },
+    {
+      "entropy": 0.1704770212993026,
+      "epoch": 9.13264722698685,
+      "grad_norm": 0.3588203489780426,
+      "learning_rate": 9.345763399590846e-07,
+      "loss": 0.07308218479156495,
+      "mean_token_accuracy": 0.9740839457511902,
+      "num_tokens": 7744145.0,
+      "step": 4000
+    },
+    {
+      "entropy": 0.17244749169796705,
+      "epoch": 9.246998284734135,
+      "grad_norm": 0.5301705598831177,
+      "learning_rate": 7.06708373127894e-07,
+      "loss": 0.0754227352142334,
+      "mean_token_accuracy": 0.9735957351326943,
+      "num_tokens": 7842944.0,
+      "step": 4050
+    },
+    {
+      "entropy": 0.17769238892942668,
+      "epoch": 9.361349342481418,
+      "grad_norm": 0.700737476348877,
+      "learning_rate": 5.101536210393669e-07,
+      "loss": 0.07621361255645752,
+      "mean_token_accuracy": 0.9725926449894905,
+      "num_tokens": 7937715.0,
+      "step": 4100
+    },
+    {
+      "entropy": 0.18282185826450587,
+      "epoch": 9.475700400228702,
+      "grad_norm": 0.46041715145111084,
+      "learning_rate": 3.4522413973552227e-07,
+      "loss": 0.07864448070526123,
+      "mean_token_accuracy": 0.9719916397333145,
+      "num_tokens": 8031402.0,
+      "step": 4150
+    },
+    {
+      "entropy": 0.1700876032561064,
+      "epoch": 9.590051457975987,
+      "grad_norm": 1.1421316862106323,
+      "learning_rate": 2.121817760585211e-07,
+      "loss": 0.07469216346740723,
+      "mean_token_accuracy": 0.9736909645795823,
+      "num_tokens": 8127966.0,
+      "step": 4200
+    },
+    {
+      "entropy": 0.1747031044960022,
+      "epoch": 9.70440251572327,
+      "grad_norm": 0.7286482453346252,
+      "learning_rate": 1.1123775193501412e-07,
+      "loss": 0.07706663608551026,
+      "mean_token_accuracy": 0.9727976828813553,
+      "num_tokens": 8223027.0,
+      "step": 4250
+    },
+    {
+      "entropy": 0.17586638078093528,
+      "epoch": 9.818753573470554,
+      "grad_norm": 0.6001836061477661,
+      "learning_rate": 4.255232903406704e-08,
+      "loss": 0.07670443534851074,
+      "mean_token_accuracy": 0.9725228261947632,
+      "num_tokens": 8318641.0,
+      "step": 4300
+    },
+    {
+      "entropy": 0.1639021010696888,
+      "epoch": 9.933104631217839,
+      "grad_norm": 0.7681242823600769,
+      "learning_rate": 6.23455433107618e-09,
+      "loss": 0.07487247467041015,
+      "mean_token_accuracy": 0.9744946631789208,
+      "num_tokens": 8419263.0,
+      "step": 4350
+    },
+    {
+      "epoch": 10.0,
+      "eval_entropy": 0.23274677067342467,
+      "eval_loss": 1.155958652496338,
+      "eval_mean_token_accuracy": 0.835232106003448,
+      "eval_num_tokens": 8476060.0,
+      "eval_runtime": 74.6637,
+      "eval_samples_per_second": 14.585,
+      "eval_steps_per_second": 1.835,
+      "step": 4380
+    }
+  ],
+  "logging_steps": 50,
+  "max_steps": 4380,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 10,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 3.6043878529489306e+17,
+  "train_batch_size": 4,
+  "trial_name": null,
+  "trial_params": null
+}

systematicity_original_Swedish/Qwen3.5-4B-Base_systematicity_splits_original_features_train_systematicity_splits_original_features_test2/checkpoint-876/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

systematicity_original_Swedish/Qwen3.5-4B-Base_systematicity_splits_original_features_train_systematicity_splits_original_features_test2/checkpoint-876/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": 256,
+  "lora_bias": false,
+  "lora_dropout": 0.0859087223648311,
+  "megatron_config": null,
+  "megatron_core": "megatron.core",
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "peft_version": "0.18.1",
+  "qalora_group_size": 16,
+  "r": 128,
+  "rank_pattern": {},
+  "revision": null,
+  "target_modules": [
+    "q_proj",
+    "k_proj",
+    "down_proj",
+    "v_proj",
+    "gate_proj",
+    "o_proj",
+    "up_proj"
+  ],
+  "target_parameters": null,
+  "task_type": "CAUSAL_LM",
+  "trainable_token_indices": null,
+  "use_dora": false,
+  "use_qalora": false,
+  "use_rslora": false
+}

systematicity_original_Swedish/Qwen3.5-4B-Base_systematicity_splits_original_features_train_systematicity_splits_original_features_test2/checkpoint-876/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 %}

systematicity_original_Swedish/Qwen3.5-4B-Base_systematicity_splits_original_features_train_systematicity_splits_original_features_test2/checkpoint-876/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|>"
+}

systematicity_original_Swedish/Qwen3.5-4B-Base_systematicity_splits_original_features_train_systematicity_splits_original_features_test2/checkpoint-876/trainer_state.json

@@ -0,0 +1,226 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 2.0,
+  "eval_steps": 500,
+  "global_step": 876,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "entropy": 2.00145958840847,
+      "epoch": 0.11435105774728416,
+      "grad_norm": 2.497065782546997,
+      "learning_rate": 4.571087888923681e-06,
+      "loss": 1.7145748901367188,
+      "mean_token_accuracy": 0.6452348600327968,
+      "num_tokens": 96586.0,
+      "step": 50
+    },
+    {
+      "entropy": 0.8970040166378022,
+      "epoch": 0.22870211549456831,
+      "grad_norm": 1.9262803792953491,
+      "learning_rate": 9.23546328578458e-06,
+      "loss": 0.7682270050048828,
+      "mean_token_accuracy": 0.7993282404541969,
+      "num_tokens": 190720.0,
+      "step": 100
+    },
+    {
+      "entropy": 0.7573070672154426,
+      "epoch": 0.34305317324185247,
+      "grad_norm": 1.519248366355896,
+      "learning_rate": 1.389983868264548e-05,
+      "loss": 0.6529589080810547,
+      "mean_token_accuracy": 0.8195512276887894,
+      "num_tokens": 288515.0,
+      "step": 150
+    },
+    {
+      "entropy": 0.7158778738975525,
+      "epoch": 0.45740423098913663,
+      "grad_norm": 1.4797775745391846,
+      "learning_rate": 1.8564214079506377e-05,
+      "loss": 0.6193486022949218,
+      "mean_token_accuracy": 0.827950005531311,
+      "num_tokens": 387874.0,
+      "step": 200
+    },
+    {
+      "entropy": 0.6906554877758027,
+      "epoch": 0.5717552887364208,
+      "grad_norm": 1.5661845207214355,
+      "learning_rate": 2.322858947636728e-05,
+      "loss": 0.5904607009887696,
+      "mean_token_accuracy": 0.8340386778116227,
+      "num_tokens": 487865.0,
+      "step": 250
+    },
+    {
+      "entropy": 0.6806062492728233,
+      "epoch": 0.6861063464837049,
+      "grad_norm": 1.276515245437622,
+      "learning_rate": 2.7892964873228177e-05,
+      "loss": 0.5840428924560547,
+      "mean_token_accuracy": 0.8354081481695175,
+      "num_tokens": 584269.0,
+      "step": 300
+    },
+    {
+      "entropy": 0.6750492942333222,
+      "epoch": 0.8004574042309891,
+      "grad_norm": 1.3048511743545532,
+      "learning_rate": 3.2557340270089076e-05,
+      "loss": 0.5731664657592773,
+      "mean_token_accuracy": 0.8381186512112617,
+      "num_tokens": 676067.0,
+      "step": 350
+    },
+    {
+      "entropy": 0.652309263497591,
+      "epoch": 0.9148084619782733,
+      "grad_norm": 1.1547592878341675,
+      "learning_rate": 3.722171566694997e-05,
+      "loss": 0.5572821807861328,
+      "mean_token_accuracy": 0.8416853401064873,
+      "num_tokens": 774009.0,
+      "step": 400
+    },
+    {
+      "epoch": 1.0,
+      "eval_entropy": 0.5965869100424495,
+      "eval_loss": 0.5746821761131287,
+      "eval_mean_token_accuracy": 0.8380154971658749,
+      "eval_num_tokens": 847606.0,
+      "eval_runtime": 75.0294,
+      "eval_samples_per_second": 14.514,
+      "eval_steps_per_second": 1.826,
+      "step": 438
+    },
+    {
+      "entropy": 0.6367223400452415,
+      "epoch": 1.0274442538593482,
+      "grad_norm": 0.936708390712738,
+      "learning_rate": 4.085914344568524e-05,
+      "loss": 0.5404758834838868,
+      "mean_token_accuracy": 0.846109490104133,
+      "num_tokens": 870214.0,
+      "step": 450
+    },
+    {
+      "entropy": 0.5927853919565678,
+      "epoch": 1.1417953116066324,
+      "grad_norm": 1.0234040021896362,
+      "learning_rate": 4.0835791756497073e-05,
+      "loss": 0.5002776718139649,
+      "mean_token_accuracy": 0.8537301248311997,
+      "num_tokens": 971584.0,
+      "step": 500
+    },
+    {
+      "entropy": 0.5995605960488319,
+      "epoch": 1.2561463693539165,
+      "grad_norm": 0.9836112856864929,
+      "learning_rate": 4.0780043181954356e-05,
+      "loss": 0.5042121124267578,
+      "mean_token_accuracy": 0.8527141407132148,
+      "num_tokens": 1063906.0,
+      "step": 550
+    },
+    {
+      "entropy": 0.5905857393145562,
+      "epoch": 1.3704974271012007,
+      "grad_norm": 1.0268890857696533,
+      "learning_rate": 4.0691986230115715e-05,
+      "loss": 0.5004239654541016,
+      "mean_token_accuracy": 0.8539833968877792,
+      "num_tokens": 1159282.0,
+      "step": 600
+    },
+    {
+      "entropy": 0.6022135411202908,
+      "epoch": 1.4848484848484849,
+      "grad_norm": 0.9739264845848083,
+      "learning_rate": 4.0571760702759385e-05,
+      "loss": 0.4968523788452148,
+      "mean_token_accuracy": 0.854298751950264,
+      "num_tokens": 1254998.0,
+      "step": 650
+    },
+    {
+      "entropy": 0.5831877724826335,
+      "epoch": 1.599199542595769,
+      "grad_norm": 0.8632996082305908,
+      "learning_rate": 4.0419557473429846e-05,
+      "loss": 0.49389530181884767,
+      "mean_token_accuracy": 0.8568908262252808,
+      "num_tokens": 1351877.0,
+      "step": 700
+    },
+    {
+      "entropy": 0.5858544714748859,
+      "epoch": 1.7135506003430532,
+      "grad_norm": 0.8835417628288269,
+      "learning_rate": 4.023561818440143e-05,
+      "loss": 0.49105472564697267,
+      "mean_token_accuracy": 0.8568279257416725,
+      "num_tokens": 1448801.0,
+      "step": 750
+    },
+    {
+      "entropy": 0.587700717896223,
+      "epoch": 1.8279016580903373,
+      "grad_norm": 0.8575327396392822,
+      "learning_rate": 4.002023486304002e-05,
+      "loss": 0.4949332046508789,
+      "mean_token_accuracy": 0.8551343831419945,
+      "num_tokens": 1547040.0,
+      "step": 800
+    },
+    {
+      "entropy": 0.5782938550412655,
+      "epoch": 1.9422527158376215,
+      "grad_norm": 0.7499766945838928,
+      "learning_rate": 3.9773749458171914e-05,
+      "loss": 0.4910758972167969,
+      "mean_token_accuracy": 0.8579927015304566,
+      "num_tokens": 1646444.0,
+      "step": 850
+    },
+    {
+      "epoch": 2.0,
+      "eval_entropy": 0.5206899273134497,
+      "eval_loss": 0.5700324177742004,
+      "eval_mean_token_accuracy": 0.8452296922676754,
+      "eval_num_tokens": 1695212.0,
+      "eval_runtime": 74.7024,
+      "eval_samples_per_second": 14.578,
+      "eval_steps_per_second": 1.834,
+      "step": 876
+    }
+  ],
+  "logging_steps": 50,
+  "max_steps": 4380,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 10,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 7.189387055737651e+16,
+  "train_batch_size": 4,
+  "trial_name": null,
+  "trial_params": null
+}