gemma4-e4b-odia8k-sft/step-1600-bf16

.gitattributes

@@ -43,3 +43,4 @@ gemma4-e4b-odia8k/warmup/tokenizer.json filter=lfs diff=lfs merge=lfs -text
 gemma4-e4b-odia8k/warmup-v2/tokenizer.json filter=lfs diff=lfs merge=lfs -text
 gemma4-e4b-odia8k-sft/step-800-bf16/tokenizer.json filter=lfs diff=lfs merge=lfs -text
 gemma4-e4b-odia8k-sft/checkpoint-1000/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+gemma4-e4b-odia8k-sft/step-1600-bf16/tokenizer.json filter=lfs diff=lfs merge=lfs -text

gemma4-e4b-odia8k-sft/step-1600-bf16/chat_template.jinja

@@ -0,0 +1,386 @@
+{#
+  Template: Google Gemma 4 Canonical Chat Template
+  Author: Google Gemma Engineering Team
+  Published: 2026-07-09
+  Context: Fixed tool-calling loops, turn closures, and thinking content-ordering.
+#}
+{%- macro format_parameters(properties, required, filter_keys=false) -%}
+    {%- set standard_keys = ['description', 'type', 'properties', 'required', 'nullable'] -%}
+    {%- set ns = namespace(found_first=false) -%}
+    {%- for key, value in properties | dictsort -%}
+        {%- set add_comma = false -%}
+        {%- if not filter_keys or key not in standard_keys -%}
+            {%- if ns.found_first %},{% endif -%}
+            {%- set ns.found_first = true -%}
+            {{ key }}:{
+            {%- if value['description'] -%}
+                description:<|"|>{{ value['description'] }}<|"|>
+                {%- set add_comma = true -%}
+            {%- endif -%}
+            {%- if value['type'] | upper == 'STRING' -%}
+                {%- if value['enum'] -%}
+                    {%- if add_comma %},{%- else -%} {%- set add_comma = true -%} {% endif -%}
+                    enum:{{ format_argument(value['enum']) }}
+                {%- endif -%}
+            {%- elif value['type'] | upper == 'ARRAY' -%}
+                {%- if value['items'] is mapping and value['items'] -%}
+                    {%- if add_comma %},{%- else -%} {%- set add_comma = true -%} {% endif -%}
+                    items:{
+                    {%- set ns_items = namespace(found_first=false) -%}
+                    {%- for item_key, item_value in value['items'] | dictsort -%}
+                        {%- if item_value is not none -%}
+                            {%- if ns_items.found_first %},{% endif -%}
+                            {%- set ns_items.found_first = true -%}
+                            {%- if item_key == 'properties' -%}
+                                properties:{
+                                {%- if item_value is mapping -%}
+                                    {{- format_parameters(item_value, value['items']['required'] | default([])) -}}
+                                {%- endif -%}
+                                }
+                            {%- elif item_key == 'required' -%}
+                                required:[
+                                {%- for req_item in item_value -%}
+                                    <|"|>{{- req_item -}}<|"|>
+                                    {%- if not loop.last %},{% endif -%}
+                                {%- endfor -%}
+                                ]
+                            {%- elif item_key == 'type' -%}
+                                {%- if item_value is string -%}
+                                    type:{{ format_argument(item_value | upper) }}
+                                {%- else -%}
+                                    type:{{ format_argument(item_value | map('upper') | list) }}
+                                {%- endif -%}
+                            {%- else -%}
+                                {{ item_key }}:{{ format_argument(item_value) }}
+                            {%- endif -%}
+                        {%- endif -%}
+                    {%- endfor -%}
+                    }
+                {%- endif -%}
+            {%- endif -%}
+            {%- if value['nullable'] %}
+                {%- if add_comma %},{%- else -%} {%- set add_comma = true -%} {% endif -%}
+                nullable:true
+            {%- endif -%}
+            {%- if value['type'] | upper == 'OBJECT' -%}
+                {%- if value['properties'] is defined and value['properties'] is mapping -%}
+                    {%- if add_comma %},{%- else -%} {%- set add_comma = true -%} {% endif -%}
+                    properties:{
+                    {{- format_parameters(value['properties'], value['required'] | default([])) -}}
+                    }
+                {%- elif value is mapping -%}
+                    {%- if add_comma %},{%- else -%} {%- set add_comma = true -%} {% endif -%}
+                    properties:{
+                    {{- format_parameters(value, value['required'] | default([]), filter_keys=true) -}}
+                    }
+                {%- endif -%}
+                {%- if value['required'] -%}
+                    {%- if add_comma %},{%- else -%} {%- set add_comma = true -%} {% endif -%}
+                    required:[
+                    {%- for item in value['required'] | default([]) -%}
+                        <|"|>{{- item -}}<|"|>
+                        {%- if not loop.last %},{% endif -%}
+                    {%- endfor -%}
+                    ]
+                {%- endif -%}
+            {%- endif -%}
+            {%- if add_comma %},{%- else -%} {%- set add_comma = true -%} {% endif -%}
+            type:<|"|>{{ value['type'] | upper }}<|"|>}
+        {%- endif -%}
+    {%- endfor -%}
+{%- endmacro -%}
+{%- macro format_function_declaration(tool_data) -%}
+    declaration:{{- tool_data['function']['name'] -}}{description:<|"|>{{- tool_data['function']['description'] -}}<|"|>
+    {%- set params = tool_data['function']['parameters'] -%}
+    {%- if params -%}
+        ,parameters:{
+        {%- if params['properties'] -%}
+            properties:{ {{- format_parameters(params['properties'], params['required']) -}} },
+        {%- endif -%}
+        {%- if params['required'] -%}
+            required:[
+            {%- for item in params['required'] -%}
+                <|"|>{{- item -}}<|"|>
+                {{- ',' if not loop.last -}}
+            {%- endfor -%}
+            ],
+        {%- endif -%}
+        {%- if params['type'] -%}
+            type:<|"|>{{- params['type'] | upper -}}<|"|>}
+        {%- endif -%}
+    {%- endif -%}
+    {%- if 'response' in tool_data['function'] -%}
+        {%- set response_declaration = tool_data['function']['response'] -%}
+        ,response:{
+        {%- if response_declaration['description'] -%}
+            description:<|"|>{{- response_declaration['description'] -}}<|"|>,
+        {%- endif -%}
+        {%- if response_declaration['type'] | upper == 'OBJECT' -%}
+            type:<|"|>{{- response_declaration['type'] | upper -}}<|"|>}
+        {%- endif -%}
+    {%- endif -%}
+    }
+{%- endmacro -%}
+{%- macro format_argument(argument, escape_keys=True) -%}
+    {%- if argument is none -%}
+        {{- 'null' -}}
+    {%- elif argument is string -%}
+        {{- '<|"|>' + argument + '<|"|>' -}}
+    {%- elif argument is boolean -%}
+        {{- 'true' if argument else 'false' -}}
+    {%- elif argument is mapping -%}
+        {{- '{' -}}
+        {%- set ns = namespace(found_first=false) -%}
+        {%- for key, value in argument | dictsort -%}
+            {%- if ns.found_first %},{% endif -%}
+            {%- set ns.found_first = true -%}
+            {%- if escape_keys -%}
+                {{- '<|"|>' + key + '<|"|>' -}}
+            {%- else -%}
+                {{- key -}}
+            {%- endif -%}
+            :{{- format_argument(value, escape_keys=escape_keys) -}}
+        {%- endfor -%}
+        {{- '}' -}}
+    {%- elif argument is sequence -%}
+        {{- '[' -}}
+        {%- for item in argument -%}
+            {{- format_argument(item, escape_keys=escape_keys) -}}
+            {%- if not loop.last %},{% endif -%}
+        {%- endfor -%}
+        {{- ']' -}}
+    {%- else -%}
+        {{- argument -}}
+    {%- endif -%}
+{%- endmacro -%}
+{%- macro strip_thinking(text) -%}
+    {%- set ns = namespace(result='') -%}
+    {%- for part in text.split('<channel|>') -%}
+        {%- if '<|channel>' in part -%}
+            {%- set ns.result = ns.result + part.split('<|channel>')[0] -%}
+        {%- else -%}
+            {%- set ns.result = ns.result + part -%}
+        {%- endif -%}
+    {%- endfor -%}
+    {{- ns.result | trim -}}
+{%- endmacro -%}
+
+{%- macro format_tool_response_block(tool_name, response) -%}
+    {{- '<|tool_response>' -}}
+    {%- if response is mapping -%}
+        {{- 'response:' + tool_name + '{' -}}
+        {%- for key, value in response | dictsort -%}
+            {{- key -}}:{{- format_argument(value, escape_keys=False) -}}
+            {%- if not loop.last %},{% endif -%}
+        {%- endfor -%}
+        {{- '}' -}}
+    {%- else -%}
+        {{- 'response:' + tool_name + '{value:' + format_argument(response, escape_keys=False) + '}' -}}
+    {%- endif -%}
+    {{- '<tool_response|>' -}}
+{%- endmacro -%}
+
+{#- ===== SETUP ===== -#}
+{%- set ns = namespace(prev_message_type=None, prev_non_tool_role=None) -%}
+{%- set loop_messages = messages -%}
+{%- set enable_thinking = enable_thinking | default(false) -%}
+{%- set preserve_thinking = preserve_thinking | default(false) -%}
+{{- bos_token -}}
+{#- Handle System/Tool Definitions Block -#}
+{%- if enable_thinking or tools or (messages and messages[0]['role'] in ['system', 'developer']) -%}
+    {{- '<|turn>system\n' -}}
+    {#- Inject Thinking token at the very top of the FIRST system turn -#}
+    {%- if enable_thinking -%}
+        {{- '<|think|>\n' -}}
+        {%- set ns.prev_message_type = 'think' -%}
+    {%- endif -%}
+    {%- if messages and messages[0]['role'] in ['system', 'developer'] -%}
+        {%- if messages[0]['content'] is string -%}
+            {{- messages[0]['content'] | trim -}}
+        {%- elif messages[0]['content'] is sequence -%}
+            {%- for item in messages[0]['content'] -%}
+                {{- item['text'] | trim + ' '-}}
+            {%- endfor -%}
+        {%- endif -%}
+        {%- set loop_messages = messages[1:] -%}
+    {%- endif -%}
+    {%- if tools -%}
+        {%- for tool in tools %}
+            {{- '<|tool>' -}}
+            {{- format_function_declaration(tool) | trim -}}
+            {{- '<tool|>' -}}
+        {%- endfor %}
+        {%- set ns.prev_message_type = 'tool' -%}
+    {%- endif -%}
+    {{- '<turn|>\n' -}}
+{%- endif %}
+
+{#- Pre-scan: find last user message index for reasoning guard -#}
+{%- set ns_turn = namespace(last_user_idx=-1) -%}
+{%- for i in range(loop_messages | length) -%}
+    {%- if loop_messages[i]['role'] == 'user' -%}
+        {%- set ns_turn.last_user_idx = i -%}
+    {%- endif -%}
+{%- endfor -%}
+
+{#- Loop through messages -#}
+{%- for message in loop_messages -%}
+    {%- if message['role'] != 'tool' -%}
+    {%- set ns.prev_message_type = None -%}
+    {%- set role = 'model' if message['role'] == 'assistant' else message['role'] -%}
+    {#- Detect continuation using tracked state — O(1) instead of O(n) backward scan -#}
+    {%- set continue_same_model_turn = (role == 'model' and ns.prev_non_tool_role == 'assistant') -%}
+    {%- if not continue_same_model_turn -%}
+        {{- '<|turn>' + role + '\n' }}
+    {%- endif -%}
+
+    {#- Render reasoning/reasoning_content as thinking channel -#}
+    {%- set thinking_text = message.get('reasoning') or message.get('reasoning_content') -%}
+    {%- set thinking_gate = (loop.index0 > ns_turn.last_user_idx) or (preserve_thinking and message.get('tool_calls')) -%}
+    {%- if thinking_text and thinking_gate -%}
+        {{- '<|channel>thought\n' + thinking_text + '\n<channel|>' -}}
+    {%- endif -%}
+
+            {%- if message.get('tool_calls') -%}
+                {%- for tool_call in message.get('tool_calls') -%}
+                    {%- set function = tool_call['function'] -%}
+                    {{- '<|tool_call>call:' + function['name'] + '{' -}}
+                    {%- if function['arguments'] is mapping -%}
+                        {%- set ns_args = namespace(found_first=false) -%}
+                        {%- for key, value in function['arguments'] | dictsort -%}
+                            {%- if ns_args.found_first %},{% endif -%}
+                            {%- set ns_args.found_first = true -%}
+                            {{- key -}}:{{- format_argument(value, escape_keys=False) -}}
+                        {%- endfor -%}
+                    {%- elif function['arguments'] is none -%}
+                    {%- else -%}
+                        {{- raise_exception(
+                            "chat_template: tool_calls[].function.arguments must be a "
+                            "JSON object (mapping), not a string. Deserialize arguments "
+                            "before passing to the template."
+                        ) -}}
+                    {%- endif -%}
+                    {{- '}<tool_call|>' -}}
+                {%- endfor -%}
+                {%- set ns.prev_message_type = 'tool_call' -%}
+            {%- endif -%}
+
+            {%- set ns_tr_out = namespace(flag=false) -%}
+            {%- if message.get('tool_responses') -%}
+                {#- Legacy: tool_responses embedded on the assistant message (Google/Gemma native) -#}
+                {%- for tool_response in message.get('tool_responses') -%}
+                    {{- format_tool_response_block(tool_response['name'] | default('unknown', true), tool_response['response']) -}}
+                    {%- set ns_tr_out.flag = true -%}
+                    {%- set ns.prev_message_type = 'tool_response' -%}
+                {%- endfor -%}
+            {%- elif message.get('tool_calls') -%}
+                {#- OpenAI Chat Completions: forward-scan consecutive role:tool messages -#}
+                {%- set ns_tool_scan = namespace(stopped=false) -%}
+                {%- for k in range(loop.index0 + 1, loop_messages | length) -%}
+                    {%- if ns_tool_scan.stopped -%}
+                    {%- elif loop_messages[k]['role'] != 'tool' -%}
+                        {%- set ns_tool_scan.stopped = true -%}
+                    {%- else -%}
+                        {%- set follow = loop_messages[k] -%}
+                        {#- Resolve tool_call_id to function name -#}
+                        {%- set ns_tname = namespace(name=follow.get('name') or 'unknown') -%}
+                        {%- for tc in message.get('tool_calls') -%}
+                            {%- if tc.get('id') == follow.get('tool_call_id') -%}
+                                {%- set ns_tname.name = tc['function']['name'] -%}
+                            {%- endif -%}
+                        {%- endfor -%}
+                        {#- Handle content as string or content-parts array -#}
+                        {%- set tool_body = follow.get('content') -%}
+                        {%- if tool_body is string -%}
+                            {{- format_tool_response_block(ns_tname.name, tool_body) -}}
+                        {%- elif tool_body is sequence and tool_body is not string -%}
+                            {%- set ns_txt = namespace(s='') -%}
+                            {%- for part in tool_body -%}
+                                {%- if part.get('type') == 'text' -%}
+                                    {%- set ns_txt.s = ns_txt.s + (part.get('text') | default('')) -%}
+                                {%- endif -%}
+                            {%- endfor -%}
+                            {{- format_tool_response_block(ns_tname.name, ns_txt.s) -}}
+                            {%- for part in tool_body -%}
+                                {%- if part.get('type') in ['image', 'image_url'] -%}
+                                    {{- '<|image|>' -}}
+                                {%- elif part.get('type') in ['audio', 'input_audio'] -%}
+                                    {{- '<|audio|>' -}}
+                                {%- elif part.get('type') == 'video' -%}
+                                    {{- '<|video|>' -}}
+                                {%- endif -%}
+                            {%- endfor -%}
+                        {%- else -%}
+                            {{- format_tool_response_block(ns_tname.name, tool_body) -}}
+                        {%- endif -%}
+                        {%- set ns_tr_out.flag = true -%}
+                        {%- set ns.prev_message_type = 'tool_response' -%}
+                    {%- endif -%}
+                {%- endfor -%}
+            {%- endif -%}
+
+            {%- set captured_content -%}
+            {%- if message.get('content') is string -%}
+                {%- if role == 'model' -%}
+                    {{- strip_thinking(message['content']) -}}
+                {%- else -%}
+                    {{- message['content'] | trim -}}
+                {%- endif -%}
+            {%- elif message.get('content') is sequence -%}
+                {%- for item in message['content'] -%}
+                    {%- if item.get('type') == 'text' -%}
+                        {%- if role == 'model' -%}
+                            {{- strip_thinking(item['text']) -}}
+                        {%- else -%}
+                            {{- item['text'] | trim -}}
+                        {%- endif -%}
+                    {%- elif item.get('type') in ['image', 'image_url'] -%}
+                        {{- '<|image|>' -}}
+                    {%- elif item.get('type') in ['audio', 'input_audio'] -%}
+                        {{- '<|audio|>' -}}
+                    {%- elif item.get('type') == 'video' -%}
+                        {{- '<|video|>' -}}
+                    {%- endif -%}
+                {%- endfor -%}
+            {%- endif -%}
+            {%- endset -%}
+
+            {{- captured_content -}}
+            {%- set has_content = captured_content | trim | length > 0 -%}
+
+        {#- Forward-scan: find next non-tool message role for continuation detection -#}
+        {%- set next_nt = namespace(role=None, found=false) -%}
+        {%- for j in range(loop.index0 + 1, loop_messages | length) -%}
+            {%- if not next_nt.found -%}
+                {%- if loop_messages[j]['role'] != 'tool' -%}
+                    {%- set next_nt.role = loop_messages[j]['role'] -%}
+                    {%- set next_nt.found = true -%}
+                {%- endif -%}
+            {%- endif -%}
+        {%- endfor -%}
+
+        {%- set continues_into_next = (
+            role == 'model'
+            and next_nt.role == 'assistant'
+            and (not message.get('tool_calls') or ns_tr_out.flag)
+        ) -%}
+
+        {%- if ns.prev_message_type == 'tool_call' and not ns_tr_out.flag -%}
+            {{- '<|tool_response>' -}}
+        {%- elif continues_into_next -%}
+        {%- elif not (ns_tr_out.flag and not has_content and not next_nt.found) -%}
+            {{- '<turn|>\n' -}}
+        {%- endif -%}
+
+    {#- Track previous non-tool role for next iteration (avoids O(n) backward scan) -#}
+    {%- set ns.prev_non_tool_role = message['role'] -%}
+    {%- endif -%}
+{%- endfor -%}
+
+{%- if add_generation_prompt -%}
+    {%- if ns.prev_message_type != 'tool_response' and ns.prev_message_type != 'tool_call' -%}
+        {{- '<|turn>model\n' -}}
+    {%- elif ns.prev_message_type == 'tool_response' and enable_thinking -%}
+        {{- '<|channel>thought\n' -}}
+    {%- endif -%}
+{%- endif -%}

gemma4-e4b-odia8k-sft/step-1600-bf16/config.json

@@ -0,0 +1,117 @@
+{
+  "architectures": [
+    "Gemma4ForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "attention_k_eq_v": false,
+  "bos_token_id": 2,
+  "dtype": "float32",
+  "enable_moe_block": false,
+  "eos_token_id": 1,
+  "expert_intermediate_size": null,
+  "final_logit_softcapping": 30.0,
+  "head_dim": 256,
+  "hidden_activation": "gelu_pytorch_tanh",
+  "hidden_size": 2560,
+  "hidden_size_per_layer_input": 256,
+  "initializer_range": 0.02,
+  "intermediate_size": 10240,
+  "layer_types": [
+    "sliding_attention",
+    "sliding_attention",
+    "sliding_attention",
+    "sliding_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "sliding_attention",
+    "sliding_attention",
+    "sliding_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "sliding_attention",
+    "sliding_attention",
+    "sliding_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "sliding_attention",
+    "sliding_attention",
+    "sliding_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "sliding_attention",
+    "sliding_attention",
+    "sliding_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "sliding_attention",
+    "sliding_attention",
+    "sliding_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "sliding_attention",
+    "sliding_attention",
+    "sliding_attention",
+    "sliding_attention",
+    "full_attention"
+  ],
+  "max_position_embeddings": 131072,
+  "model_type": "gemma4_text",
+  "moe_intermediate_size": null,
+  "num_attention_heads": 8,
+  "num_experts": null,
+  "num_hidden_layers": 42,
+  "num_key_value_heads": 2,
+  "num_kv_shared_layers": 18,
+  "pad_token_id": 0,
+  "per_layer_config": {
+    "05": {
+      "head_dim": 512
+    },
+    "11": {
+      "head_dim": 512
+    },
+    "17": {
+      "head_dim": 512
+    },
+    "23": {
+      "head_dim": 512
+    },
+    "29": {
+      "head_dim": 512
+    },
+    "35": {
+      "head_dim": 512
+    },
+    "41": {
+      "head_dim": 512
+    }
+  },
+  "rms_norm_eps": 1e-06,
+  "rope_parameters": {
+    "full_attention": {
+      "partial_rotary_factor": 0.25,
+      "rope_theta": 1000000.0,
+      "rope_type": "proportional"
+    },
+    "sliding_attention": {
+      "rope_theta": 10000.0,
+      "rope_type": "default"
+    }
+  },
+  "sliding_window": 512,
+  "tie_word_embeddings": true,
+  "top_k_experts": null,
+  "transformers_version": "5.17.0",
+  "use_bidirectional_attention": null,
+  "use_cache": false,
+  "use_double_wide_mlp": false,
+  "vocab_size": 161407,
+  "vocab_size_per_layer_input": 161407
+}

gemma4-e4b-odia8k-sft/step-1600-bf16/generation_config.json

@@ -0,0 +1,14 @@
+{
+  "bos_token_id": 2,
+  "do_sample": true,
+  "eos_token_id": [
+    1,
+    17,
+    10
+  ],
+  "pad_token_id": 0,
+  "temperature": 1.0,
+  "top_k": 64,
+  "top_p": 0.95,
+  "transformers_version": "5.17.0"
+}

gemma4-e4b-odia8k-sft/step-1600-bf16/model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c5905d5a942dcd9f49439e36a8fcbf00a758f3276129a5176b6acd5d0c12be48
+size 13070486780

gemma4-e4b-odia8k-sft/step-1600-bf16/tokenizer.json

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f891c4968a939cafec8a7d4b4652d01c0e43ae2e5f04cf39e2d7e8ea3670037b
+size 22835663

gemma4-e4b-odia8k-sft/step-1600-bf16/tokenizer_config.json

@@ -0,0 +1,126 @@
+{
+  "backend": "tokenizers",
+  "bos_token": "<bos>",
+  "eoc_token": "<channel|>",
+  "eos_token": "<eos>",
+  "eot_token": "<turn|>",
+  "escape_token": "<|\"|>",
+  "etc_token": "<tool_call|>",
+  "etd_token": "<tool|>",
+  "etr_token": "<tool_response|>",
+  "is_local": true,
+  "local_files_only": false,
+  "mask_token": "<mask>",
+  "model_max_length": 1000000000000000019884624838656,
+  "model_specific_special_tokens": {
+    "eoc_token": "<channel|>",
+    "eot_token": "<turn|>",
+    "escape_token": "<|\"|>",
+    "etc_token": "<tool_call|>",
+    "etd_token": "<tool|>",
+    "etr_token": "<tool_response|>",
+    "soc_token": "<|channel>",
+    "sot_token": "<|turn>",
+    "stc_token": "<|tool_call>",
+    "std_token": "<|tool>",
+    "str_token": "<|tool_response>",
+    "think_token": "<|think|>"
+  },
+  "pad_token": "<pad>",
+  "padding_side": "left",
+  "response_schema": {
+    "properties": {
+      "content": {
+        "type": "string"
+      },
+      "role": {
+        "const": "assistant"
+      },
+      "thinking": {
+        "type": "string"
+      },
+      "tool_calls": {
+        "items": {
+          "properties": {
+            "function": {
+              "properties": {
+                "arguments": {
+                  "additionalProperties": {},
+                  "type": "object",
+                  "x-parser": "gemma4-tool-call"
+                },
+                "name": {
+                  "type": "string"
+                }
+              },
+              "type": "object",
+              "x-regex": "call\\:(?P<name>\\w+)(?P<arguments>\\{.*\\})"
+            },
+            "type": {
+              "const": "function"
+            }
+          },
+          "type": "object"
+        },
+        "type": "array",
+        "x-regex-iterator": "<\\|tool_call>(.*?)<tool_call\\|>"
+      }
+    },
+    "type": "object",
+    "x-regex": "(\\<\\|channel\\>thought\\n(?P<thinking>.*?)\\<channel\\|\\>)?(?P<tool_calls>\\<\\|tool_call\\>.*\\<tool_call\\|\\>)?(?P<content>(?:(?!\\<turn\\|\\>)(?!\\<\\|tool_response\\>).)+)?(?:\\<turn\\|\\>|\\<\\|tool_response\\>)?"
+  },
+  "response_template": {
+    "defaults": {
+      "role": "assistant"
+    },
+    "fields": {
+      "content": {
+        "close": [
+          "<turn|>",
+          "<|tool_response>",
+          "<eos>"
+        ],
+        "content": "text"
+      },
+      "thinking": {
+        "close": "<channel|>",
+        "content": "text",
+        "open": "<|channel>thought\n"
+      },
+      "tool_calls": {
+        "close": "<tool_call|>",
+        "content": "json",
+        "content_args": {
+          "string_delims": [
+            [
+              "<|\"|>",
+              "<|\"|>"
+            ]
+          ],
+          "unquoted_keys": true
+        },
+        "open_pattern": "<\\|tool_call>call:(?P<name>\\w+)",
+        "repeats": true,
+        "transform": {
+          "function": {
+            "arguments": "{content}",
+            "name": "{name}"
+          },
+          "type": "function"
+        }
+      }
+    },
+    "start_anchor": [
+      "<|turn>model\n",
+      "<tool_response|>"
+    ]
+  },
+  "soc_token": "<|channel>",
+  "sot_token": "<|turn>",
+  "stc_token": "<|tool_call>",
+  "std_token": "<|tool>",
+  "str_token": "<|tool_response>",
+  "think_token": "<|think|>",
+  "tokenizer_class": "GemmaTokenizer",
+  "unk_token": "<unk>"
+}

gemma4-e4b-odia8k-sft/step-1600-bf16/trainer_state.json

@@ -0,0 +1 @@
+{"step": 1600, "epoch": 0.7814407814407814, "log_history": [{"eval_sft_loss": 0.8725103139877319, "eval_sft_runtime": 31.4838, "eval_sft_samples_per_second": 49.359, "eval_sft_steps_per_second": 6.194, "eval_entropy": 0.8000412881374359, "eval_num_tokens": 0.0, "eval_mean_token_accuracy": 0.8057849238213804, "epoch": 0, "step": 0}, {"loss": 0.9166790843009949, "grad_norm": 6.180274963378906, "learning_rate": 0.0, "padded_tokens": 16384, "real_tokens": 16384, "pad_ratio": 1.0, "tokens_per_sec": 3999, "mfu": 0.1064, "entropy": 0.907211184501648, "num_tokens": 16384.0, "mean_token_accuracy": 0.784228652715683, "epoch": 0.0004884004884004884, "step": 1}, {"loss": 1.0300446616278753, "grad_norm": 6.055756092071533, "learning_rate": 1.4516129032258066e-06, "padded_tokens": 163791, "real_tokens": 163791, "pad_ratio": 1.0, "tokens_per_sec": 4579, "mfu": 0.1219, "entropy": 0.9543175283405516, "num_tokens": 163791.0, "mean_token_accuracy": 0.7707291907734342, "epoch": 0.004884004884004884, "step": 10}, {"loss": 1.0082554817199707, "grad_norm": 6.844918727874756, "learning_rate": 3.0645161290322584e-06, "padded_tokens": 327499, "real_tokens": 327499, "pad_ratio": 1.0, "tokens_per_sec": 4613, "mfu": 0.1228, "entropy": 1.0132723167538642, "num_tokens": 327499.0, "mean_token_accuracy": 0.7696516260504722, "epoch": 0.009768009768009768, "step": 20}, {"loss": 0.847509765625, "grad_norm": 7.4658284187316895, "learning_rate": 4.67741935483871e-06, "padded_tokens": 488875, "real_tokens": 488875, "pad_ratio": 1.0, "tokens_per_sec": 4590, "mfu": 0.1222, "entropy": 0.8202234834432602, "num_tokens": 488875.0, "mean_token_accuracy": 0.7938304170966148, "epoch": 0.014652014652014652, "step": 30}, {"loss": 0.8819092750549317, "grad_norm": 4.292705059051514, "learning_rate": 6.290322580645162e-06, "padded_tokens": 652606, "real_tokens": 652606, "pad_ratio": 1.0, "tokens_per_sec": 4636, "mfu": 0.1234, "entropy": 0.8659455761313438, "num_tokens": 652606.0, "mean_token_accuracy": 0.7904599353671073, "epoch": 0.019536019536019536, "step": 40}, {"loss": 0.830943489074707, "grad_norm": 4.519834518432617, "learning_rate": 7.903225806451613e-06, "padded_tokens": 816280, "real_tokens": 816280, "pad_ratio": 1.0, "tokens_per_sec": 4575, "mfu": 0.1217, "entropy": 0.8024719037115574, "num_tokens": 816280.0, "mean_token_accuracy": 0.7990727290511132, "epoch": 0.02442002442002442, "step": 50}, {"loss": 0.9408459663391113, "grad_norm": 10.839374542236328, "learning_rate": 9.516129032258065e-06, "padded_tokens": 974796, "real_tokens": 974796, "pad_ratio": 1.0, "tokens_per_sec": 4622, "mfu": 0.123, "entropy": 0.9375784993171692, "num_tokens": 974796.0, "mean_token_accuracy": 0.769383916258812, "epoch": 0.029304029304029304, "step": 60}, {"loss": 0.8366189956665039, "grad_norm": 5.091339588165283, "learning_rate": 9.999724123048889e-06, "padded_tokens": 1137309, "real_tokens": 1137309, "pad_ratio": 1.0, "tokens_per_sec": 4615, "mfu": 0.1228, "entropy": 0.8245442263782025, "num_tokens": 1137309.0, "mean_token_accuracy": 0.7960970371961593, "epoch": 0.03418803418803419, "step": 70}, {"loss": 0.8763234138488769, "grad_norm": 6.620410919189453, "learning_rate": 9.998372970431915e-06, "padded_tokens": 1301063, "real_tokens": 1301063, "pad_ratio": 1.0, "tokens_per_sec": 4697, "mfu": 0.125, "entropy": 0.8566777631640434, "num_tokens": 1301063.0, "mean_token_accuracy": 0.7919274553656578, "epoch": 0.03907203907203907, "step": 80}, {"loss": 0.8516076087951661, "grad_norm": 3.9351017475128174, "learning_rate": 9.995896208543021e-06, "padded_tokens": 1461418, "real_tokens": 1461418, "pad_ratio": 1.0, "tokens_per_sec": 4562, "mfu": 0.1214, "entropy": 0.8303296439349651, "num_tokens": 1461418.0, "mean_token_accuracy": 0.7961921468377113, "epoch": 0.04395604395604396, "step": 90}, {"loss": 0.7481894969940186, "grad_norm": 18.0093936920166, "learning_rate": 9.992294457132099e-06, "padded_tokens": 1622427, "real_tokens": 1622427, "pad_ratio": 1.0, "tokens_per_sec": 4574, "mfu": 0.1217, "entropy": 0.7623457759618759, "num_tokens": 1622427.0, "mean_token_accuracy": 0.8037400856614113, "epoch": 0.04884004884004884, "step": 100}, {"loss": 0.7629738807678222, "grad_norm": 4.2351975440979, "learning_rate": 9.987568617450513e-06, "padded_tokens": 1786192, "real_tokens": 1786192, "pad_ratio": 1.0, "tokens_per_sec": 4673, "mfu": 0.1244, "entropy": 0.7575705341994763, "num_tokens": 1786192.0, "mean_token_accuracy": 0.8051830351352691, "epoch": 0.05372405372405373, "step": 110}, {"loss": 0.7552663326263428, "grad_norm": 4.743204593658447, "learning_rate": 9.98171987202559e-06, "padded_tokens": 1949898, "real_tokens": 1949898, "pad_ratio": 1.0, "tokens_per_sec": 4677, "mfu": 0.1245, "entropy": 0.7665151834487915, "num_tokens": 1949898.0, "mean_token_accuracy": 0.8020474046468735, "epoch": 0.05860805860805861, "step": 120}, {"loss": 0.7985593795776367, "grad_norm": 3.1051957607269287, "learning_rate": 9.974749684364717e-06, "padded_tokens": 2113634, "real_tokens": 2113634, "pad_ratio": 1.0, "tokens_per_sec": 4674, "mfu": 0.1244, "entropy": 0.787737463414669, "num_tokens": 2113634.0, "mean_token_accuracy": 0.7990755841135979, "epoch": 0.06349206349206349, "step": 130}, {"loss": 0.7226286888122558, "grad_norm": 4.946564674377441, "learning_rate": 9.96665979858914e-06, "padded_tokens": 2277329, "real_tokens": 2277329, "pad_ratio": 1.0, "tokens_per_sec": 4690, "mfu": 0.1248, "entropy": 0.7353189773857594, "num_tokens": 2277329.0, "mean_token_accuracy": 0.8105007097125053, "epoch": 0.06837606837606838, "step": 140}, {"loss": 0.7685462474822998, "grad_norm": 3.6530373096466064, "learning_rate": 9.957452238997533e-06, "padded_tokens": 2439196, "real_tokens": 2439196, "pad_ratio": 1.0, "tokens_per_sec": 4606, "mfu": 0.1226, "entropy": 0.7713561035692692, "num_tokens": 2439196.0, "mean_token_accuracy": 0.802637341618538, "epoch": 0.07326007326007326, "step": 150}, {"loss": 0.721370267868042, "grad_norm": 3.4713127613067627, "learning_rate": 9.947129309559466e-06, "padded_tokens": 2600412, "real_tokens": 2600412, "pad_ratio": 1.0, "tokens_per_sec": 4636, "mfu": 0.1234, "entropy": 0.7158238537609577, "num_tokens": 2600412.0, "mean_token_accuracy": 0.8083542183041572, "epoch": 0.07814407814407814, "step": 160}, {"loss": 0.7015998363494873, "grad_norm": 3.831228733062744, "learning_rate": 9.935693593338898e-06, "padded_tokens": 2760489, "real_tokens": 2760489, "pad_ratio": 1.0, "tokens_per_sec": 4679, "mfu": 0.1245, "entropy": 0.7128293827176094, "num_tokens": 2760489.0, "mean_token_accuracy": 0.8086942970752716, "epoch": 0.08302808302808302, "step": 170}, {"loss": 0.7088464736938477, "grad_norm": 3.1233787536621094, "learning_rate": 9.923147951847821e-06, "padded_tokens": 2919786, "real_tokens": 2919786, "pad_ratio": 1.0, "tokens_per_sec": 4568, "mfu": 0.1216, "entropy": 0.7418126873672009, "num_tokens": 2919786.0, "mean_token_accuracy": 0.807001693546772, "epoch": 0.08791208791208792, "step": 180}, {"loss": 0.7075233459472656, "grad_norm": 3.62237286567688, "learning_rate": 9.909495524330245e-06, "padded_tokens": 3083598, "real_tokens": 3083598, "pad_ratio": 1.0, "tokens_per_sec": 4680, "mfu": 0.1246, "entropy": 0.6956375062465667, "num_tokens": 3083598.0, "mean_token_accuracy": 0.8109288826584816, "epoch": 0.0927960927960928, "step": 190}, {"loss": 0.752910041809082, "grad_norm": 2.8915576934814453, "learning_rate": 9.894739726976664e-06, "padded_tokens": 3247296, "real_tokens": 3247296, "pad_ratio": 1.0, "tokens_per_sec": 4656, "mfu": 0.1239, "entropy": 0.7588844999670983, "num_tokens": 3247296.0, "mean_token_accuracy": 0.8028429508209228, "epoch": 0.09768009768009768, "step": 200}, {"loss": 0.6615881443023681, "grad_norm": 3.9374582767486572, "learning_rate": 9.878884252069252e-06, "padded_tokens": 3411068, "real_tokens": 3411068, "pad_ratio": 1.0, "tokens_per_sec": 4664, "mfu": 0.1241, "entropy": 0.6475411631166935, "num_tokens": 3411068.0, "mean_token_accuracy": 0.8222755298018456, "epoch": 0.10256410256410256, "step": 210}, {"loss": 0.6269526958465577, "grad_norm": 2.905413866043091, "learning_rate": 9.861933067057947e-06, "padded_tokens": 3570883, "real_tokens": 3570883, "pad_ratio": 1.0, "tokens_per_sec": 4574, "mfu": 0.1217, "entropy": 0.6572307974100113, "num_tokens": 3570883.0, "mean_token_accuracy": 0.8206150755286217, "epoch": 0.10744810744810745, "step": 220}, {"loss": 0.6993788242340088, "grad_norm": 3.099003791809082, "learning_rate": 9.843890413567696e-06, "padded_tokens": 3734625, "real_tokens": 3734625, "pad_ratio": 1.0, "tokens_per_sec": 4643, "mfu": 0.1236, "entropy": 0.7123510889708996, "num_tokens": 3734625.0, "mean_token_accuracy": 0.8082438617944717, "epoch": 0.11233211233211234, "step": 230}, {"loss": 0.6573081970214844, "grad_norm": 3.235427141189575, "learning_rate": 9.824760806337092e-06, "padded_tokens": 3898369, "real_tokens": 3898369, "pad_ratio": 1.0, "tokens_per_sec": 4632, "mfu": 0.1233, "entropy": 0.6529365874826908, "num_tokens": 3898369.0, "mean_token_accuracy": 0.8208374693989754, "epoch": 0.11721611721611722, "step": 240}, {"loss": 0.6546825885772705, "grad_norm": 3.356523275375366, "learning_rate": 9.804549032088672e-06, "padded_tokens": 4062052, "real_tokens": 4062052, "pad_ratio": 1.0, "tokens_per_sec": 4656, "mfu": 0.1239, "entropy": 0.6679666399955749, "num_tokens": 4062052.0, "mean_token_accuracy": 0.8180171996355057, "epoch": 0.1221001221001221, "step": 250}, {"loss": 0.7024047374725342, "grad_norm": 3.602299928665161, "learning_rate": 9.78326014833115e-06, "padded_tokens": 4225766, "real_tokens": 4225766, "pad_ratio": 1.0, "tokens_per_sec": 4650, "mfu": 0.1237, "entropy": 0.7213010050356388, "num_tokens": 4225766.0, "mean_token_accuracy": 0.807885679602623, "epoch": 0.12698412698412698, "step": 260}, {"loss": 0.7629943370819092, "grad_norm": 3.4527945518493652, "learning_rate": 9.760899482093897e-06, "padded_tokens": 4389498, "real_tokens": 4389498, "pad_ratio": 1.0, "tokens_per_sec": 4680, "mfu": 0.1245, "entropy": 0.7667133763432503, "num_tokens": 4389498.0, "mean_token_accuracy": 0.7963322460651397, "epoch": 0.13186813186813187, "step": 270}, {"loss": 0.6417510509490967, "grad_norm": 4.2771711349487305, "learning_rate": 9.737472628593988e-06, "padded_tokens": 4551840, "real_tokens": 4551840, "pad_ratio": 1.0, "tokens_per_sec": 4682, "mfu": 0.1246, "entropy": 0.6432803243398666, "num_tokens": 4551840.0, "mean_token_accuracy": 0.8218334585428237, "epoch": 0.13675213675213677, "step": 280}, {"loss": 0.5923994541168213, "grad_norm": 3.5229737758636475, "learning_rate": 9.712985449836126e-06, "padded_tokens": 4715463, "real_tokens": 4715463, "pad_ratio": 1.0, "tokens_per_sec": 4688, "mfu": 0.1247, "entropy": 0.6043940663337708, "num_tokens": 4715463.0, "mean_token_accuracy": 0.8307585686445236, "epoch": 0.14163614163614163, "step": 290}, {"loss": 0.6646195888519287, "grad_norm": 3.641875982284546, "learning_rate": 9.687444073145818e-06, "padded_tokens": 4877387, "real_tokens": 4877387, "pad_ratio": 1.0, "tokens_per_sec": 4673, "mfu": 0.1244, "entropy": 0.6739590600132942, "num_tokens": 4877387.0, "mean_token_accuracy": 0.8156840965151787, "epoch": 0.14652014652014653, "step": 300}, {"loss": 0.6261611461639405, "grad_norm": 4.39824104309082, "learning_rate": 9.660854889636169e-06, "padded_tokens": 5040051, "real_tokens": 5040051, "pad_ratio": 1.0, "tokens_per_sec": 4658, "mfu": 0.124, "entropy": 0.6299219764769077, "num_tokens": 5040051.0, "mean_token_accuracy": 0.8235479891300201, "epoch": 0.1514041514041514, "step": 310}, {"loss": 0.6570065975189209, "grad_norm": 13.30288028717041, "learning_rate": 9.633224552608644e-06, "padded_tokens": 5201555, "real_tokens": 5201555, "pad_ratio": 1.0, "tokens_per_sec": 4658, "mfu": 0.1239, "entropy": 0.66151417568326, "num_tokens": 5201555.0, "mean_token_accuracy": 0.8184865832328796, "epoch": 0.1562881562881563, "step": 320}, {"loss": 0.5671521186828613, "grad_norm": 3.6645472049713135, "learning_rate": 9.604559975888252e-06, "padded_tokens": 5362984, "real_tokens": 5362984, "pad_ratio": 1.0, "tokens_per_sec": 4594, "mfu": 0.1223, "entropy": 0.5947534717619419, "num_tokens": 5362984.0, "mean_token_accuracy": 0.8349431365728378, "epoch": 0.16117216117216118, "step": 330}, {"loss": 0.6113387107849121, "grad_norm": 3.883409023284912, "learning_rate": 9.574868332093527e-06, "padded_tokens": 5518571, "real_tokens": 5518571, "pad_ratio": 1.0, "tokens_per_sec": 4592, "mfu": 0.1222, "entropy": 0.6451925709843636, "num_tokens": 5518571.0, "mean_token_accuracy": 0.8210435152053833, "epoch": 0.16605616605616605, "step": 340}, {"loss": 0.6932660102844238, "grad_norm": 3.9356448650360107, "learning_rate": 9.544157050841748e-06, "padded_tokens": 5679532, "real_tokens": 5679532, "pad_ratio": 1.0, "tokens_per_sec": 4656, "mfu": 0.1239, "entropy": 0.7324018262326717, "num_tokens": 5679532.0, "mean_token_accuracy": 0.8109753012657166, "epoch": 0.17094017094017094, "step": 350}, {"loss": 0.6577122211456299, "grad_norm": 4.00785493850708, "learning_rate": 9.512433816889863e-06, "padded_tokens": 5843249, "real_tokens": 5843249, "pad_ratio": 1.0, "tokens_per_sec": 4698, "mfu": 0.125, "entropy": 0.6644346199929714, "num_tokens": 5843249.0, "mean_token_accuracy": 0.8203870460391045, "epoch": 0.17582417582417584, "step": 360}, {"loss": 0.5888994216918946, "grad_norm": 3.355776309967041, "learning_rate": 9.479706568211556e-06, "padded_tokens": 6005042, "real_tokens": 6005042, "pad_ratio": 1.0, "tokens_per_sec": 4675, "mfu": 0.1244, "entropy": 0.593878085911274, "num_tokens": 6005042.0, "mean_token_accuracy": 0.8365008592605591, "epoch": 0.1807081807081807, "step": 370}, {"loss": 0.7303317070007325, "grad_norm": 2.9303042888641357, "learning_rate": 9.445983494010968e-06, "padded_tokens": 6168615, "real_tokens": 6168615, "pad_ratio": 1.0, "tokens_per_sec": 4663, "mfu": 0.1241, "entropy": 0.7287182964384555, "num_tokens": 6168615.0, "mean_token_accuracy": 0.8066214621067047, "epoch": 0.1855921855921856, "step": 380}, {"loss": 0.6534792423248291, "grad_norm": 3.1694533824920654, "learning_rate": 9.411273032673547e-06, "padded_tokens": 6332152, "real_tokens": 6332152, "pad_ratio": 1.0, "tokens_per_sec": 4655, "mfu": 0.1239, "entropy": 0.6594708636403084, "num_tokens": 6332152.0, "mean_token_accuracy": 0.8205794423818589, "epoch": 0.19047619047619047, "step": 390}, {"loss": 0.6406374931335449, "grad_norm": 2.885751962661743, "learning_rate": 9.375583869654533e-06, "padded_tokens": 6495832, "real_tokens": 6495832, "pad_ratio": 1.0, "tokens_per_sec": 4674, "mfu": 0.1244, "entropy": 0.6551136463880539, "num_tokens": 6495832.0, "mean_token_accuracy": 0.8215438410639763, "epoch": 0.19536019536019536, "step": 400}, {"loss": 0.8071249961853028, "grad_norm": 4.370692253112793, "learning_rate": 9.338924935305653e-06, "padded_tokens": 6657835, "real_tokens": 6657835, "pad_ratio": 1.0, "tokens_per_sec": 4645, "mfu": 0.1236, "entropy": 0.7954380415380001, "num_tokens": 6657835.0, "mean_token_accuracy": 0.7979190677404404, "epoch": 0.20024420024420025, "step": 410}, {"loss": 0.6451890945434571, "grad_norm": 3.4769346714019775, "learning_rate": 9.301305402640497e-06, "padded_tokens": 6820553, "real_tokens": 6820553, "pad_ratio": 1.0, "tokens_per_sec": 4704, "mfu": 0.1252, "entropy": 0.6543322511017322, "num_tokens": 6820553.0, "mean_token_accuracy": 0.8259319975972176, "epoch": 0.20512820512820512, "step": 420}, {"loss": 0.7163309097290039, "grad_norm": 3.9925377368927, "learning_rate": 9.262734685039204e-06, "padded_tokens": 6984080, "real_tokens": 6984080, "pad_ratio": 1.0, "tokens_per_sec": 4704, "mfu": 0.1252, "entropy": 0.7107013992965221, "num_tokens": 6984080.0, "mean_token_accuracy": 0.8089245140552521, "epoch": 0.21001221001221002, "step": 430}, {"loss": 0.6212490558624267, "grad_norm": 4.2436370849609375, "learning_rate": 9.223222433892985e-06, "padded_tokens": 7143826, "real_tokens": 7143826, "pad_ratio": 1.0, "tokens_per_sec": 4635, "mfu": 0.1234, "entropy": 0.6681630335748195, "num_tokens": 7143826.0, "mean_token_accuracy": 0.821677677333355, "epoch": 0.2148962148962149, "step": 440}, {"loss": 0.6437687873840332, "grad_norm": 3.550196886062622, "learning_rate": 9.182778536189099e-06, "padded_tokens": 7307525, "real_tokens": 7307525, "pad_ratio": 1.0, "tokens_per_sec": 4692, "mfu": 0.1249, "entropy": 0.6335056573152542, "num_tokens": 7307525.0, "mean_token_accuracy": 0.8239563822746276, "epoch": 0.21978021978021978, "step": 450}, {"loss": 0.7123647689819336, "grad_norm": 5.130570411682129, "learning_rate": 9.141413112036873e-06, "padded_tokens": 7469365, "real_tokens": 7469365, "pad_ratio": 1.0, "tokens_per_sec": 4655, "mfu": 0.1239, "entropy": 0.7359176993370056, "num_tokens": 7469365.0, "mean_token_accuracy": 0.8122776925563813, "epoch": 0.22466422466422467, "step": 460}, {"loss": 0.6472006797790527, "grad_norm": 4.005089282989502, "learning_rate": 9.099136512135384e-06, "padded_tokens": 7629978, "real_tokens": 7629978, "pad_ratio": 1.0, "tokens_per_sec": 4624, "mfu": 0.1231, "entropy": 0.6620700307190418, "num_tokens": 7629978.0, "mean_token_accuracy": 0.8225894585251808, "epoch": 0.22954822954822954, "step": 470}, {"loss": 0.6563704967498779, "grad_norm": 3.481660842895508, "learning_rate": 9.055959315183454e-06, "padded_tokens": 7793651, "real_tokens": 7793651, "pad_ratio": 1.0, "tokens_per_sec": 4670, "mfu": 0.1243, "entropy": 0.6633002400398255, "num_tokens": 7793651.0, "mean_token_accuracy": 0.8234314531087875, "epoch": 0.23443223443223443, "step": 480}, {"loss": 0.6544801712036132, "grad_norm": 3.909770965576172, "learning_rate": 9.011892325232578e-06, "padded_tokens": 7957354, "real_tokens": 7957354, "pad_ratio": 1.0, "tokens_per_sec": 4670, "mfu": 0.1243, "entropy": 0.6508172109723092, "num_tokens": 7957354.0, "mean_token_accuracy": 0.8246287748217582, "epoch": 0.23931623931623933, "step": 490}, {"loss": 0.6520584106445313, "grad_norm": 3.169631004333496, "learning_rate": 8.966946568983479e-06, "padded_tokens": 8120978, "real_tokens": 8120978, "pad_ratio": 1.0, "tokens_per_sec": 4646, "mfu": 0.1236, "entropy": 0.679977161437273, "num_tokens": 8120978.0, "mean_token_accuracy": 0.821967813372612, "epoch": 0.2442002442002442, "step": 500}, {"eval_sft_loss": 0.7248380780220032, "eval_sft_runtime": 21.8997, "eval_sft_samples_per_second": 70.96, "eval_sft_steps_per_second": 8.904, "eval_entropy": 0.574930835015995, "eval_num_tokens": 8120978.0, "eval_mean_token_accuracy": 0.8304713306967745, "epoch": 0.2442002442002442, "step": 500}, {"loss": 0.727435302734375, "grad_norm": 4.70125150680542, "learning_rate": 8.921133293026934e-06, "padded_tokens": 8277234, "real_tokens": 8277234, "pad_ratio": 1.0, "tokens_per_sec": 12577, "mfu": 0.3347, "entropy": 0.7520171493291855, "num_tokens": 8277234.0, "mean_token_accuracy": 0.8097755521535873, "epoch": 0.2490842490842491, "step": 510}, {"loss": 0.6937822818756103, "grad_norm": 3.9462060928344727, "learning_rate": 8.87446396102959e-06, "padded_tokens": 8439777, "real_tokens": 8439777, "pad_ratio": 1.0, "tokens_per_sec": 4640, "mfu": 0.1235, "entropy": 0.6893045477569103, "num_tokens": 8439777.0, "mean_token_accuracy": 0.817629761993885, "epoch": 0.25396825396825395, "step": 520}, {"loss": 0.6624418258666992, "grad_norm": 3.4985203742980957, "learning_rate": 8.826950250865464e-06, "padded_tokens": 8600992, "real_tokens": 8600992, "pad_ratio": 1.0, "tokens_per_sec": 4644, "mfu": 0.1236, "entropy": 0.6704959169030189, "num_tokens": 8600992.0, "mean_token_accuracy": 0.8251426666975021, "epoch": 0.2588522588522589, "step": 530}, {"loss": 0.7216361045837403, "grad_norm": 4.304588794708252, "learning_rate": 8.778604051693821e-06, "padded_tokens": 8764277, "real_tokens": 8764277, "pad_ratio": 1.0, "tokens_per_sec": 4675, "mfu": 0.1244, "entropy": 0.7397279486060142, "num_tokens": 8764277.0, "mean_token_accuracy": 0.8149169027805329, "epoch": 0.26373626373626374, "step": 540}, {"loss": 0.7096574306488037, "grad_norm": 2.822355270385742, "learning_rate": 8.72943746098422e-06, "padded_tokens": 8925256, "real_tokens": 8925256, "pad_ratio": 1.0, "tokens_per_sec": 4596, "mfu": 0.1223, "entropy": 0.7226120330393314, "num_tokens": 8925256.0, "mean_token_accuracy": 0.8170015409588813, "epoch": 0.2686202686202686, "step": 550}, {"loss": 0.760306453704834, "grad_norm": 4.395480632781982, "learning_rate": 8.679462781489392e-06, "padded_tokens": 9087052, "real_tokens": 9087052, "pad_ratio": 1.0, "tokens_per_sec": 4685, "mfu": 0.1247, "entropy": 0.7532184183597564, "num_tokens": 9087052.0, "mean_token_accuracy": 0.8069549798965454, "epoch": 0.27350427350427353, "step": 560}, {"loss": 0.739990520477295, "grad_norm": 3.8643014430999756, "learning_rate": 8.628692518166792e-06, "padded_tokens": 9242509, "real_tokens": 9242509, "pad_ratio": 1.0, "tokens_per_sec": 4566, "mfu": 0.1215, "entropy": 0.7645664907991886, "num_tokens": 9242509.0, "mean_token_accuracy": 0.8115659862756729, "epoch": 0.2783882783882784, "step": 570}, {"loss": 0.6930558681488037, "grad_norm": 3.1767678260803223, "learning_rate": 8.577139375049528e-06, "padded_tokens": 9406180, "real_tokens": 9406180, "pad_ratio": 1.0, "tokens_per_sec": 4687, "mfu": 0.1247, "entropy": 0.7190431781113148, "num_tokens": 9406180.0, "mean_token_accuracy": 0.8181145042181015, "epoch": 0.28327228327228327, "step": 580}, {"loss": 0.7542586326599121, "grad_norm": 4.147907733917236, "learning_rate": 8.524816252067473e-06, "padded_tokens": 9569910, "real_tokens": 9569910, "pad_ratio": 1.0, "tokens_per_sec": 4658, "mfu": 0.124, "entropy": 0.7348876148462296, "num_tokens": 9569910.0, "mean_token_accuracy": 0.8123106166720391, "epoch": 0.28815628815628813, "step": 590}, {"loss": 0.8089179039001465, "grad_norm": 3.067108154296875, "learning_rate": 8.471736241819388e-06, "padded_tokens": 9731336, "real_tokens": 9731336, "pad_ratio": 1.0, "tokens_per_sec": 4698, "mfu": 0.125, "entropy": 0.7945016272366047, "num_tokens": 9731336.0, "mean_token_accuracy": 0.803807869553566, "epoch": 0.29304029304029305, "step": 600}, {"loss": 0.8160871505737305, "grad_norm": 5.330318450927734, "learning_rate": 8.417912626296802e-06, "padded_tokens": 9895076, "real_tokens": 9895076, "pad_ratio": 1.0, "tokens_per_sec": 4699, "mfu": 0.1251, "entropy": 0.8077732779085636, "num_tokens": 9895076.0, "mean_token_accuracy": 0.8074929356575012, "epoch": 0.2979242979242979, "step": 610}, {"loss": 0.7779770851135254, "grad_norm": 4.3128533363342285, "learning_rate": 8.363358873560519e-06, "padded_tokens": 10058598, "real_tokens": 10058598, "pad_ratio": 1.0, "tokens_per_sec": 4634, "mfu": 0.1233, "entropy": 0.7563504435122013, "num_tokens": 10058598.0, "mean_token_accuracy": 0.8126975119113922, "epoch": 0.3028083028083028, "step": 620}, {"loss": 0.7383967399597168, "grad_norm": 4.023048400878906, "learning_rate": 8.308088634370555e-06, "padded_tokens": 10219484, "real_tokens": 10219484, "pad_ratio": 1.0, "tokens_per_sec": 4635, "mfu": 0.1233, "entropy": 0.7442464418709278, "num_tokens": 10219484.0, "mean_token_accuracy": 0.8139302998781204, "epoch": 0.3076923076923077, "step": 630}, {"loss": 0.789900827407837, "grad_norm": 3.1395456790924072, "learning_rate": 8.252115738770378e-06, "padded_tokens": 10383135, "real_tokens": 10383135, "pad_ratio": 1.0, "tokens_per_sec": 4668, "mfu": 0.1242, "entropy": 0.777092257887125, "num_tokens": 10383135.0, "mean_token_accuracy": 0.811117199063301, "epoch": 0.3125763125763126, "step": 640}, {"loss": 0.8453151702880859, "grad_norm": 4.099407196044922, "learning_rate": 8.195454192626255e-06, "padded_tokens": 10543218, "real_tokens": 10543218, "pad_ratio": 1.0, "tokens_per_sec": 4630, "mfu": 0.1232, "entropy": 0.8601422473788262, "num_tokens": 10543218.0, "mean_token_accuracy": 0.7903747275471688, "epoch": 0.31746031746031744, "step": 650}, {"loss": 0.71726393699646, "grad_norm": 2.695868968963623, "learning_rate": 8.13811817412264e-06, "padded_tokens": 10703834, "real_tokens": 10703834, "pad_ratio": 1.0, "tokens_per_sec": 4694, "mfu": 0.1249, "entropy": 0.7290932253003121, "num_tokens": 10703834.0, "mean_token_accuracy": 0.8187746480107307, "epoch": 0.32234432234432236, "step": 660}, {"loss": 0.772478723526001, "grad_norm": 2.946150779724121, "learning_rate": 8.08012203021441e-06, "padded_tokens": 10867588, "real_tokens": 10867588, "pad_ratio": 1.0, "tokens_per_sec": 4660, "mfu": 0.124, "entropy": 0.7647079937160015, "num_tokens": 10867588.0, "mean_token_accuracy": 0.8088037326931954, "epoch": 0.32722832722832723, "step": 670}, {"loss": 0.7808233261108398, "grad_norm": 2.6063427925109863, "learning_rate": 8.0214802730369e-06, "padded_tokens": 11031328, "real_tokens": 11031328, "pad_ratio": 1.0, "tokens_per_sec": 4653, "mfu": 0.1238, "entropy": 0.7707127839326858, "num_tokens": 11031328.0, "mean_token_accuracy": 0.808801707625389, "epoch": 0.3321123321123321, "step": 680}, {"loss": 0.835174560546875, "grad_norm": 3.1612846851348877, "learning_rate": 7.962207576274577e-06, "padded_tokens": 11193530, "real_tokens": 11193530, "pad_ratio": 1.0, "tokens_per_sec": 4644, "mfu": 0.1236, "entropy": 0.8287145465612411, "num_tokens": 11193530.0, "mean_token_accuracy": 0.8004722639918327, "epoch": 0.336996336996337, "step": 690}, {"loss": 0.6832539558410644, "grad_norm": 3.216346263885498, "learning_rate": 7.902318771489321e-06, "padded_tokens": 11357245, "real_tokens": 11357245, "pad_ratio": 1.0, "tokens_per_sec": 4666, "mfu": 0.1242, "entropy": 0.6850744426250458, "num_tokens": 11357245.0, "mean_token_accuracy": 0.8271145477890969, "epoch": 0.3418803418803419, "step": 700}, {"loss": 0.75770583152771, "grad_norm": 3.247758626937866, "learning_rate": 7.841828844409166e-06, "padded_tokens": 11520993, "real_tokens": 11520993, "pad_ratio": 1.0, "tokens_per_sec": 4674, "mfu": 0.1244, "entropy": 0.7674289546906948, "num_tokens": 11520993.0, "mean_token_accuracy": 0.8099973872303963, "epoch": 0.34676434676434675, "step": 710}, {"loss": 0.6699800491333008, "grad_norm": 5.431860446929932, "learning_rate": 7.780752931178498e-06, "padded_tokens": 11683158, "real_tokens": 11683158, "pad_ratio": 1.0, "tokens_per_sec": 4660, "mfu": 0.124, "entropy": 0.6731983177363873, "num_tokens": 11683158.0, "mean_token_accuracy": 0.8282152220606804, "epoch": 0.3516483516483517, "step": 720}, {"loss": 0.8536211013793945, "grad_norm": 3.00593900680542, "learning_rate": 7.719106314570596e-06, "padded_tokens": 11843680, "real_tokens": 11843680, "pad_ratio": 1.0, "tokens_per_sec": 4649, "mfu": 0.1237, "entropy": 0.8866537228226662, "num_tokens": 11843680.0, "mean_token_accuracy": 0.7909204736351967, "epoch": 0.35653235653235654, "step": 730}, {"loss": 0.8142979621887207, "grad_norm": 3.733219623565674, "learning_rate": 7.656904420163484e-06, "padded_tokens": 12000212, "real_tokens": 12000212, "pad_ratio": 1.0, "tokens_per_sec": 4606, "mfu": 0.1226, "entropy": 0.8073889523744583, "num_tokens": 12000212.0, "mean_token_accuracy": 0.802325963973999, "epoch": 0.3614163614163614, "step": 740}, {"loss": 0.7198453903198242, "grad_norm": 2.60439395904541, "learning_rate": 7.59416281248005e-06, "padded_tokens": 12163952, "real_tokens": 12163952, "pad_ratio": 1.0, "tokens_per_sec": 4689, "mfu": 0.1248, "entropy": 0.7184702150523663, "num_tokens": 12163952.0, "mean_token_accuracy": 0.8211049243807793, "epoch": 0.3663003663003663, "step": 750}, {"loss": 0.7502670288085938, "grad_norm": 2.9797253608703613, "learning_rate": 7.530897191093412e-06, "padded_tokens": 12327607, "real_tokens": 12327607, "pad_ratio": 1.0, "tokens_per_sec": 4719, "mfu": 0.1256, "entropy": 0.7559881910681725, "num_tokens": 12327607.0, "mean_token_accuracy": 0.8136648878455162, "epoch": 0.3711843711843712, "step": 760}, {"loss": 0.7350879192352295, "grad_norm": 4.475875377655029, "learning_rate": 7.4671233866984676e-06, "padded_tokens": 12491171, "real_tokens": 12491171, "pad_ratio": 1.0, "tokens_per_sec": 4681, "mfu": 0.1246, "entropy": 0.7312693424522877, "num_tokens": 12491171.0, "mean_token_accuracy": 0.8145410507917404, "epoch": 0.37606837606837606, "step": 770}, {"loss": 0.7910858154296875, "grad_norm": 3.178539276123047, "learning_rate": 7.402857357150645e-06, "padded_tokens": 12652916, "real_tokens": 12652916, "pad_ratio": 1.0, "tokens_per_sec": 4628, "mfu": 0.1232, "entropy": 0.7930116087198258, "num_tokens": 12652916.0, "mean_token_accuracy": 0.8099809646606445, "epoch": 0.38095238095238093, "step": 780}, {"loss": 0.7536827087402344, "grad_norm": 3.077683210372925, "learning_rate": 7.338115183472836e-06, "padded_tokens": 12816526, "real_tokens": 12816526, "pad_ratio": 1.0, "tokens_per_sec": 4678, "mfu": 0.1245, "entropy": 0.743191609531641, "num_tokens": 12816526.0, "mean_token_accuracy": 0.8156776338815689, "epoch": 0.38583638583638585, "step": 790}, {"loss": 0.749139404296875, "grad_norm": 3.494023084640503, "learning_rate": 7.272913065831519e-06, "padded_tokens": 12976626, "real_tokens": 12976626, "pad_ratio": 1.0, "tokens_per_sec": 3009, "mfu": 0.0801, "entropy": 0.7705717541277408, "num_tokens": 12976626.0, "mean_token_accuracy": 0.8080866530537605, "epoch": 0.3907203907203907, "step": 800}, {"loss": 0.747089433670044, "grad_norm": 3.127915620803833, "learning_rate": 7.20726731948304e-06, "padded_tokens": 13140385, "real_tokens": 13140385, "pad_ratio": 1.0, "tokens_per_sec": 4665, "mfu": 0.1241, "entropy": 0.7308873318135738, "num_tokens": 13140385.0, "mean_token_accuracy": 0.8181188076734542, "epoch": 0.3956043956043956, "step": 810}, {"loss": 0.682232666015625, "grad_norm": 4.352582931518555, "learning_rate": 7.141194370691124e-06, "padded_tokens": 13304111, "real_tokens": 13304111, "pad_ratio": 1.0, "tokens_per_sec": 4679, "mfu": 0.1245, "entropy": 0.6980711862444877, "num_tokens": 13304111.0, "mean_token_accuracy": 0.8251571491360664, "epoch": 0.4004884004884005, "step": 820}, {"loss": 0.7399104595184326, "grad_norm": 3.0563478469848633, "learning_rate": 7.074710752616603e-06, "padded_tokens": 13464266, "real_tokens": 13464266, "pad_ratio": 1.0, "tokens_per_sec": 4653, "mfu": 0.1238, "entropy": 0.7512718282639981, "num_tokens": 13464266.0, "mean_token_accuracy": 0.8087157145142555, "epoch": 0.4053724053724054, "step": 830}, {"loss": 0.6989696025848389, "grad_norm": 3.7711756229400635, "learning_rate": 7.0078331011803746e-06, "padded_tokens": 13624605, "real_tokens": 13624605, "pad_ratio": 1.0, "tokens_per_sec": 4623, "mfu": 0.123, "entropy": 0.7101220495998859, "num_tokens": 13624605.0, "mean_token_accuracy": 0.8235661491751671, "epoch": 0.41025641025641024, "step": 840}, {"loss": 0.7260991096496582, "grad_norm": 3.359679937362671, "learning_rate": 6.940578150900691e-06, "padded_tokens": 13784542, "real_tokens": 13784542, "pad_ratio": 1.0, "tokens_per_sec": 4620, "mfu": 0.1229, "entropy": 0.7472021333873272, "num_tokens": 13784542.0, "mean_token_accuracy": 0.8138013198971749, "epoch": 0.41514041514041516, "step": 850}, {"loss": 0.777623176574707, "grad_norm": 2.838900089263916, "learning_rate": 6.872962730705719e-06, "padded_tokens": 13944576, "real_tokens": 13944576, "pad_ratio": 1.0, "tokens_per_sec": 4585, "mfu": 0.122, "entropy": 0.7873943038284779, "num_tokens": 13944576.0, "mean_token_accuracy": 0.8073016002774238, "epoch": 0.42002442002442003, "step": 860}, {"loss": 0.7260656833648682, "grad_norm": 3.1824746131896973, "learning_rate": 6.805003759722532e-06, "padded_tokens": 14104910, "real_tokens": 14104910, "pad_ratio": 1.0, "tokens_per_sec": 4621, "mfu": 0.123, "entropy": 0.744601520895958, "num_tokens": 14104910.0, "mean_token_accuracy": 0.8144696950912476, "epoch": 0.4249084249084249, "step": 870}, {"loss": 0.7385654926300049, "grad_norm": 4.647879600524902, "learning_rate": 6.736718243043493e-06, "padded_tokens": 14265404, "real_tokens": 14265404, "pad_ratio": 1.0, "tokens_per_sec": 4672, "mfu": 0.1243, "entropy": 0.7696101561188697, "num_tokens": 14265404.0, "mean_token_accuracy": 0.8134439840912819, "epoch": 0.4297924297924298, "step": 880}, {"loss": 0.7239140033721924, "grad_norm": 2.90320086479187, "learning_rate": 6.668123267471145e-06, "padded_tokens": 14429144, "real_tokens": 14429144, "pad_ratio": 1.0, "tokens_per_sec": 4709, "mfu": 0.1253, "entropy": 0.7225788973271847, "num_tokens": 14429144.0, "mean_token_accuracy": 0.8181302577257157, "epoch": 0.4346764346764347, "step": 890}, {"loss": 0.7931922912597656, "grad_norm": 3.2771198749542236, "learning_rate": 6.599235997242652e-06, "padded_tokens": 14590028, "real_tokens": 14590028, "pad_ratio": 1.0, "tokens_per_sec": 4669, "mfu": 0.1242, "entropy": 0.7982104517519474, "num_tokens": 14590028.0, "mean_token_accuracy": 0.8069354355335235, "epoch": 0.43956043956043955, "step": 900}, {"loss": 0.756435489654541, "grad_norm": 3.1745779514312744, "learning_rate": 6.530073669734857e-06, "padded_tokens": 14752650, "real_tokens": 14752650, "pad_ratio": 1.0, "tokens_per_sec": 4685, "mfu": 0.1247, "entropy": 0.7378392890095711, "num_tokens": 14752650.0, "mean_token_accuracy": 0.8165401488542556, "epoch": 0.4444444444444444, "step": 910}, {"loss": 0.8137457847595215, "grad_norm": 4.696133613586426, "learning_rate": 6.460653591151056e-06, "padded_tokens": 14914821, "real_tokens": 14914821, "pad_ratio": 1.0, "tokens_per_sec": 4689, "mfu": 0.1248, "entropy": 0.8122925251722336, "num_tokens": 14914821.0, "mean_token_accuracy": 0.8049655050039292, "epoch": 0.44932844932844934, "step": 920}, {"loss": 0.7375215530395508, "grad_norm": 2.7642247676849365, "learning_rate": 6.390993132190527e-06, "padded_tokens": 15078572, "real_tokens": 15078572, "pad_ratio": 1.0, "tokens_per_sec": 4657, "mfu": 0.1239, "entropy": 0.7285373590886592, "num_tokens": 15078572.0, "mean_token_accuracy": 0.8170687526464462, "epoch": 0.4542124542124542, "step": 930}, {"loss": 0.6692588329315186, "grad_norm": 3.2210564613342285, "learning_rate": 6.321109723701934e-06, "padded_tokens": 15242288, "real_tokens": 15242288, "pad_ratio": 1.0, "tokens_per_sec": 4651, "mfu": 0.1238, "entropy": 0.6501925587654114, "num_tokens": 15242288.0, "mean_token_accuracy": 0.831752672791481, "epoch": 0.4590964590964591, "step": 940}, {"loss": 0.79177827835083, "grad_norm": 3.5329554080963135, "learning_rate": 6.251020852321683e-06, "padded_tokens": 15405616, "real_tokens": 15405616, "pad_ratio": 1.0, "tokens_per_sec": 4671, "mfu": 0.1243, "entropy": 0.8019728191196919, "num_tokens": 15405616.0, "mean_token_accuracy": 0.8068763077259063, "epoch": 0.463980463980464, "step": 950}, {"loss": 0.7734987735748291, "grad_norm": 3.4248783588409424, "learning_rate": 6.180744056098308e-06, "padded_tokens": 15569035, "real_tokens": 15569035, "pad_ratio": 1.0, "tokens_per_sec": 4652, "mfu": 0.1238, "entropy": 0.7989964716136455, "num_tokens": 15569035.0, "mean_token_accuracy": 0.8063551589846611, "epoch": 0.46886446886446886, "step": 960}, {"loss": 0.6993268013000489, "grad_norm": 3.5538735389709473, "learning_rate": 6.110296920104004e-06, "padded_tokens": 15727643, "real_tokens": 15727643, "pad_ratio": 1.0, "tokens_per_sec": 4579, "mfu": 0.1219, "entropy": 0.7129235923290252, "num_tokens": 15727643.0, "mean_token_accuracy": 0.8208787053823471, "epoch": 0.47374847374847373, "step": 970}, {"loss": 0.669735050201416, "grad_norm": 3.43831729888916, "learning_rate": 6.039697072034375e-06, "padded_tokens": 15891243, "real_tokens": 15891243, "pad_ratio": 1.0, "tokens_per_sec": 4657, "mfu": 0.1239, "entropy": 0.6782803863286972, "num_tokens": 15891243.0, "mean_token_accuracy": 0.8261294409632682, "epoch": 0.47863247863247865, "step": 980}, {"loss": 0.8157914161682129, "grad_norm": 3.350924253463745, "learning_rate": 5.968962177797544e-06, "padded_tokens": 16051150, "real_tokens": 16051150, "pad_ratio": 1.0, "tokens_per_sec": 4637, "mfu": 0.1234, "entropy": 0.8145887367427349, "num_tokens": 16051150.0, "mean_token_accuracy": 0.8054885670542717, "epoch": 0.4835164835164835, "step": 990}, {"loss": 0.71977219581604, "grad_norm": 3.4740190505981445, "learning_rate": 5.898109937093671e-06, "padded_tokens": 16213446, "real_tokens": 16213446, "pad_ratio": 1.0, "tokens_per_sec": 4662, "mfu": 0.1241, "entropy": 0.7433199346065521, "num_tokens": 16213446.0, "mean_token_accuracy": 0.8156566083431244, "epoch": 0.4884004884004884, "step": 1000}, {"eval_sft_loss": 0.6371726393699646, "eval_sft_runtime": 21.9408, "eval_sft_samples_per_second": 70.827, "eval_sft_steps_per_second": 8.888, "eval_entropy": 0.5799698029289541, "eval_num_tokens": 16213446.0, "eval_mean_token_accuracy": 0.8421708754042989, "epoch": 0.4884004884004884, "step": 1000}, {"loss": 0.6757226943969726, "grad_norm": 3.4513051509857178, "learning_rate": 5.827158078986046e-06, "padded_tokens": 16377125, "real_tokens": 16377125, "pad_ratio": 1.0, "tokens_per_sec": 1353, "mfu": 0.036, "entropy": 0.7068902015686035, "num_tokens": 16377125.0, "mean_token_accuracy": 0.8212860345840454, "epoch": 0.4932844932844933, "step": 1010}, {"loss": 0.7719696044921875, "grad_norm": 3.1506290435791016, "learning_rate": 5.7561243574648e-06, "padded_tokens": 16539257, "real_tokens": 16539257, "pad_ratio": 1.0, "tokens_per_sec": 4657, "mfu": 0.1239, "entropy": 0.7784990184009075, "num_tokens": 16539257.0, "mean_token_accuracy": 0.8124727442860603, "epoch": 0.4981684981684982, "step": 1020}, {"loss": 0.7044988632202148, "grad_norm": 2.6580617427825928, "learning_rate": 5.685026547004415e-06, "padded_tokens": 16699874, "real_tokens": 16699874, "pad_ratio": 1.0, "tokens_per_sec": 4602, "mfu": 0.1225, "entropy": 0.7303443044424057, "num_tokens": 16699874.0, "mean_token_accuracy": 0.8194909930229187, "epoch": 0.503052503052503, "step": 1030}, {"loss": 0.7525791168212891, "grad_norm": 3.069901466369629, "learning_rate": 5.613882438116083e-06, "padded_tokens": 16863459, "real_tokens": 16863459, "pad_ratio": 1.0, "tokens_per_sec": 4670, "mfu": 0.1243, "entropy": 0.7762688621878624, "num_tokens": 16863459.0, "mean_token_accuracy": 0.8111702963709831, "epoch": 0.5079365079365079, "step": 1040}, {"loss": 0.7036087512969971, "grad_norm": 10.60265064239502, "learning_rate": 5.542709832896054e-06, "padded_tokens": 17024664, "real_tokens": 17024664, "pad_ratio": 1.0, "tokens_per_sec": 4638, "mfu": 0.1234, "entropy": 0.714270731806755, "num_tokens": 17024664.0, "mean_token_accuracy": 0.8162231609225273, "epoch": 0.5128205128205128, "step": 1050}, {"loss": 0.7019691467285156, "grad_norm": 4.213215351104736, "learning_rate": 5.471526540571106e-06, "padded_tokens": 17186946, "real_tokens": 17186946, "pad_ratio": 1.0, "tokens_per_sec": 4661, "mfu": 0.124, "entropy": 0.7189229592680931, "num_tokens": 17186946.0, "mean_token_accuracy": 0.821591579914093, "epoch": 0.5177045177045178, "step": 1060}, {"loss": 0.7311814785003662, "grad_norm": 2.565305233001709, "learning_rate": 5.400350373042198e-06, "padded_tokens": 17347226, "real_tokens": 17347226, "pad_ratio": 1.0, "tokens_per_sec": 4619, "mfu": 0.1229, "entropy": 0.7461310960352421, "num_tokens": 17347226.0, "mean_token_accuracy": 0.8139090463519096, "epoch": 0.5225885225885226, "step": 1070}, {"loss": 0.7188891887664794, "grad_norm": 3.601593494415283, "learning_rate": 5.329199140427489e-06, "padded_tokens": 17510389, "real_tokens": 17510389, "pad_ratio": 1.0, "tokens_per_sec": 4683, "mfu": 0.1246, "entropy": 0.7347336314618588, "num_tokens": 17510389.0, "mean_token_accuracy": 0.8179189547896385, "epoch": 0.5274725274725275, "step": 1080}, {"loss": 0.7055974483489991, "grad_norm": 2.437930107116699, "learning_rate": 5.258090646605772e-06, "padded_tokens": 17670708, "real_tokens": 17670708, "pad_ratio": 1.0, "tokens_per_sec": 4628, "mfu": 0.1232, "entropy": 0.7324783727526665, "num_tokens": 17670708.0, "mean_token_accuracy": 0.816527071595192, "epoch": 0.5323565323565324, "step": 1090}, {"loss": 0.7696905136108398, "grad_norm": 3.150311231613159, "learning_rate": 5.187042684761492e-06, "padded_tokens": 17834452, "real_tokens": 17834452, "pad_ratio": 1.0, "tokens_per_sec": 4713, "mfu": 0.1254, "entropy": 0.749271722137928, "num_tokens": 17834452.0, "mean_token_accuracy": 0.8150355622172356, "epoch": 0.5372405372405372, "step": 1100}, {"loss": 0.666237735748291, "grad_norm": 2.853602886199951, "learning_rate": 5.116073032932424e-06, "padded_tokens": 17998151, "real_tokens": 17998151, "pad_ratio": 1.0, "tokens_per_sec": 4660, "mfu": 0.124, "entropy": 0.697215986251831, "num_tokens": 17998151.0, "mean_token_accuracy": 0.8268684923648835, "epoch": 0.5421245421245421, "step": 1110}, {"loss": 0.7820762634277344, "grad_norm": 3.191270351409912, "learning_rate": 5.045199449561153e-06, "padded_tokens": 18158306, "real_tokens": 18158306, "pad_ratio": 1.0, "tokens_per_sec": 4601, "mfu": 0.1224, "entropy": 0.7779399305582047, "num_tokens": 18158306.0, "mean_token_accuracy": 0.811860716342926, "epoch": 0.5470085470085471, "step": 1120}, {"loss": 0.7203388690948487, "grad_norm": 3.806466579437256, "learning_rate": 4.9744396690514555e-06, "padded_tokens": 18321919, "real_tokens": 18321919, "pad_ratio": 1.0, "tokens_per_sec": 4657, "mfu": 0.1239, "entropy": 0.7221531473100186, "num_tokens": 18321919.0, "mean_token_accuracy": 0.8214187428355217, "epoch": 0.5518925518925519, "step": 1130}, {"loss": 0.6569449424743652, "grad_norm": 3.635981559753418, "learning_rate": 4.903811397330679e-06, "padded_tokens": 18483457, "real_tokens": 18483457, "pad_ratio": 1.0, "tokens_per_sec": 4648, "mfu": 0.1237, "entropy": 0.6929212495684623, "num_tokens": 18483457.0, "mean_token_accuracy": 0.8286027908325195, "epoch": 0.5567765567765568, "step": 1140}, {"loss": 0.7053001403808594, "grad_norm": 3.299119710922241, "learning_rate": 4.833332307419278e-06, "padded_tokens": 18647125, "real_tokens": 18647125, "pad_ratio": 1.0, "tokens_per_sec": 4670, "mfu": 0.1243, "entropy": 0.6787692978978157, "num_tokens": 18647125.0, "mean_token_accuracy": 0.8236540570855141, "epoch": 0.5616605616605617, "step": 1150}, {"loss": 0.8300942420959473, "grad_norm": 3.170236587524414, "learning_rate": 4.76302003500856e-06, "padded_tokens": 18807794, "real_tokens": 18807794, "pad_ratio": 1.0, "tokens_per_sec": 4632, "mfu": 0.1233, "entropy": 0.8444305844604969, "num_tokens": 18807794.0, "mean_token_accuracy": 0.7978711187839508, "epoch": 0.5665445665445665, "step": 1160}, {"loss": 0.6927874565124512, "grad_norm": 3.3735485076904297, "learning_rate": 4.692892174047774e-06, "padded_tokens": 18971516, "real_tokens": 18971516, "pad_ratio": 1.0, "tokens_per_sec": 4678, "mfu": 0.1245, "entropy": 0.6769750691950321, "num_tokens": 18971516.0, "mean_token_accuracy": 0.8256358176469802, "epoch": 0.5714285714285714, "step": 1170}, {"loss": 0.6222124099731445, "grad_norm": 3.9072182178497314, "learning_rate": 4.622966272341653e-06, "padded_tokens": 19135164, "real_tokens": 19135164, "pad_ratio": 1.0, "tokens_per_sec": 4655, "mfu": 0.1239, "entropy": 0.630989746004343, "num_tokens": 19135164.0, "mean_token_accuracy": 0.834474328160286, "epoch": 0.5763125763125763, "step": 1180}, {"loss": 0.6317719459533692, "grad_norm": 2.985487222671509, "learning_rate": 4.553259827159511e-06, "padded_tokens": 19298872, "real_tokens": 19298872, "pad_ratio": 1.0, "tokens_per_sec": 4662, "mfu": 0.1241, "entropy": 0.6212092913687229, "num_tokens": 19298872.0, "mean_token_accuracy": 0.8369322493672371, "epoch": 0.5811965811965812, "step": 1190}, {"loss": 0.6754543781280518, "grad_norm": 2.7929580211639404, "learning_rate": 4.483790280856952e-06, "padded_tokens": 19459465, "real_tokens": 19459465, "pad_ratio": 1.0, "tokens_per_sec": 4626, "mfu": 0.1231, "entropy": 0.693233822286129, "num_tokens": 19459465.0, "mean_token_accuracy": 0.8266499415040016, "epoch": 0.5860805860805861, "step": 1200}, {"loss": 0.7076825141906739, "grad_norm": 2.735924005508423, "learning_rate": 4.414575016511359e-06, "padded_tokens": 19620796, "real_tokens": 19620796, "pad_ratio": 1.0, "tokens_per_sec": 4658, "mfu": 0.124, "entropy": 0.7080821298062802, "num_tokens": 19620796.0, "mean_token_accuracy": 0.8181044206023216, "epoch": 0.590964590964591, "step": 1210}, {"loss": 0.6213553428649903, "grad_norm": 2.6614432334899902, "learning_rate": 4.345631353572187e-06, "padded_tokens": 19784545, "real_tokens": 19784545, "pad_ratio": 1.0, "tokens_per_sec": 4622, "mfu": 0.123, "entropy": 0.6499016106128692, "num_tokens": 19784545.0, "mean_token_accuracy": 0.8347972825169563, "epoch": 0.5958485958485958, "step": 1220}, {"loss": 0.6124848842620849, "grad_norm": 2.5252950191497803, "learning_rate": 4.2769765435272e-06, "padded_tokens": 19948242, "real_tokens": 19948242, "pad_ratio": 1.0, "tokens_per_sec": 4650, "mfu": 0.1237, "entropy": 0.6189284183084964, "num_tokens": 19948242.0, "mean_token_accuracy": 0.8349811628460884, "epoch": 0.6007326007326007, "step": 1230}, {"loss": 0.7116284847259522, "grad_norm": 3.358247756958008, "learning_rate": 4.208627765585696e-06, "padded_tokens": 20108266, "real_tokens": 20108266, "pad_ratio": 1.0, "tokens_per_sec": 4633, "mfu": 0.1233, "entropy": 0.7505401507019996, "num_tokens": 20108266.0, "mean_token_accuracy": 0.8203459456562996, "epoch": 0.6056166056166056, "step": 1240}, {"loss": 0.7580090522766113, "grad_norm": 2.914788007736206, "learning_rate": 4.140602122379817e-06, "padded_tokens": 20269361, "real_tokens": 20269361, "pad_ratio": 1.0, "tokens_per_sec": 4604, "mfu": 0.1225, "entropy": 0.7594766482710839, "num_tokens": 20269361.0, "mean_token_accuracy": 0.8097841307520867, "epoch": 0.6105006105006106, "step": 1250}, {"loss": 0.6909577369689941, "grad_norm": 2.86920166015625, "learning_rate": 4.072916635685044e-06, "padded_tokens": 20429122, "real_tokens": 20429122, "pad_ratio": 1.0, "tokens_per_sec": 4631, "mfu": 0.1232, "entropy": 0.7104828670620918, "num_tokens": 20429122.0, "mean_token_accuracy": 0.823690478503704, "epoch": 0.6153846153846154, "step": 1260}, {"loss": 0.6711246490478515, "grad_norm": 2.87247371673584, "learning_rate": 4.005588242160898e-06, "padded_tokens": 20590052, "real_tokens": 20590052, "pad_ratio": 1.0, "tokens_per_sec": 4667, "mfu": 0.1242, "entropy": 0.6895471431314946, "num_tokens": 20590052.0, "mean_token_accuracy": 0.8262078821659088, "epoch": 0.6202686202686203, "step": 1270}, {"loss": 0.8149631500244141, "grad_norm": 3.452228546142578, "learning_rate": 3.938633789112951e-06, "padded_tokens": 20753676, "real_tokens": 20753676, "pad_ratio": 1.0, "tokens_per_sec": 4631, "mfu": 0.1233, "entropy": 0.8206601589918137, "num_tokens": 20753676.0, "mean_token_accuracy": 0.7989798292517662, "epoch": 0.6251526251526252, "step": 1280}, {"loss": 0.6234654426574707, "grad_norm": 4.235274314880371, "learning_rate": 3.872070030277197e-06, "padded_tokens": 20917369, "real_tokens": 20917369, "pad_ratio": 1.0, "tokens_per_sec": 4680, "mfu": 0.1245, "entropy": 0.6470661029219628, "num_tokens": 20917369.0, "mean_token_accuracy": 0.8370958864688873, "epoch": 0.63003663003663, "step": 1290}, {"loss": 0.7092997074127197, "grad_norm": 3.3838889598846436, "learning_rate": 3.8059136216278314e-06, "padded_tokens": 21075711, "real_tokens": 21075711, "pad_ratio": 1.0, "tokens_per_sec": 4623, "mfu": 0.123, "entropy": 0.7210148684680462, "num_tokens": 21075711.0, "mean_token_accuracy": 0.8194180876016617, "epoch": 0.6349206349206349, "step": 1300}, {"loss": 0.6257568836212158, "grad_norm": 2.811051368713379, "learning_rate": 3.7401811172095003e-06, "padded_tokens": 21239415, "real_tokens": 21239415, "pad_ratio": 1.0, "tokens_per_sec": 4651, "mfu": 0.1238, "entropy": 0.6537309408187866, "num_tokens": 21239415.0, "mean_token_accuracy": 0.8377915054559708, "epoch": 0.6398046398046398, "step": 1310}, {"loss": 0.6706645011901855, "grad_norm": 3.056992769241333, "learning_rate": 3.674888964995039e-06, "padded_tokens": 21399058, "real_tokens": 21399058, "pad_ratio": 1.0, "tokens_per_sec": 4629, "mfu": 0.1232, "entropy": 0.6735585682094097, "num_tokens": 21399058.0, "mean_token_accuracy": 0.8283934742212296, "epoch": 0.6446886446886447, "step": 1320}, {"loss": 0.7364355087280273, "grad_norm": 2.959742546081543, "learning_rate": 3.610053502769776e-06, "padded_tokens": 21562731, "real_tokens": 21562731, "pad_ratio": 1.0, "tokens_per_sec": 4657, "mfu": 0.1239, "entropy": 0.7594939514994621, "num_tokens": 21562731.0, "mean_token_accuracy": 0.8130943655967713, "epoch": 0.6495726495726496, "step": 1330}, {"loss": 0.7257959365844726, "grad_norm": 3.8205296993255615, "learning_rate": 3.545690954043396e-06, "padded_tokens": 21724949, "real_tokens": 21724949, "pad_ratio": 1.0, "tokens_per_sec": 4664, "mfu": 0.1241, "entropy": 0.7560117907822133, "num_tokens": 21724949.0, "mean_token_accuracy": 0.8145994007587433, "epoch": 0.6544566544566545, "step": 1340}, {"loss": 0.6341359615325928, "grad_norm": 2.883147954940796, "learning_rate": 3.4818174239903823e-06, "padded_tokens": 21888720, "real_tokens": 21888720, "pad_ratio": 1.0, "tokens_per_sec": 4697, "mfu": 0.125, "entropy": 0.639369685202837, "num_tokens": 21888720.0, "mean_token_accuracy": 0.8345712423324585, "epoch": 0.6593406593406593, "step": 1350}, {"loss": 0.7274582862854004, "grad_norm": 3.5205118656158447, "learning_rate": 3.4184488954201074e-06, "padded_tokens": 22049282, "real_tokens": 22049282, "pad_ratio": 1.0, "tokens_per_sec": 4602, "mfu": 0.1225, "entropy": 0.717702828347683, "num_tokens": 22049282.0, "mean_token_accuracy": 0.8166425496339798, "epoch": 0.6642246642246642, "step": 1360}, {"loss": 0.7400991439819335, "grad_norm": 4.021768093109131, "learning_rate": 3.355601224777507e-06, "padded_tokens": 22213021, "real_tokens": 22213021, "pad_ratio": 1.0, "tokens_per_sec": 4702, "mfu": 0.1251, "entropy": 0.7472447134554386, "num_tokens": 22213021.0, "mean_token_accuracy": 0.8173058927059174, "epoch": 0.6691086691086691, "step": 1370}, {"loss": 0.6544260501861572, "grad_norm": 2.830230712890625, "learning_rate": 3.2932901381753916e-06, "padded_tokens": 22376712, "real_tokens": 22376712, "pad_ratio": 1.0, "tokens_per_sec": 4661, "mfu": 0.124, "entropy": 0.6679035507142543, "num_tokens": 22376712.0, "mean_token_accuracy": 0.83382728099823, "epoch": 0.673992673992674, "step": 1380}, {"loss": 0.7749702453613281, "grad_norm": 2.899466037750244, "learning_rate": 3.231531227459365e-06, "padded_tokens": 22538978, "real_tokens": 22538978, "pad_ratio": 1.0, "tokens_per_sec": 4644, "mfu": 0.1236, "entropy": 0.7617383696138859, "num_tokens": 22538978.0, "mean_token_accuracy": 0.8107786789536476, "epoch": 0.6788766788766789, "step": 1390}, {"loss": 0.7829278469085693, "grad_norm": 3.289257049560547, "learning_rate": 3.1703399463063394e-06, "padded_tokens": 22700387, "real_tokens": 22700387, "pad_ratio": 1.0, "tokens_per_sec": 4692, "mfu": 0.1249, "entropy": 0.8153060056269169, "num_tokens": 22700387.0, "mean_token_accuracy": 0.8073073565959931, "epoch": 0.6837606837606838, "step": 1400}, {"loss": 0.6126925945281982, "grad_norm": 3.0706517696380615, "learning_rate": 3.1097316063576323e-06, "padded_tokens": 22861265, "real_tokens": 22861265, "pad_ratio": 1.0, "tokens_per_sec": 4657, "mfu": 0.1239, "entropy": 0.6468134805560112, "num_tokens": 22861265.0, "mean_token_accuracy": 0.8354204773902894, "epoch": 0.6886446886446886, "step": 1410}, {"loss": 0.7162126541137696, "grad_norm": 3.9369983673095703, "learning_rate": 3.049721373387591e-06, "padded_tokens": 23021901, "real_tokens": 23021901, "pad_ratio": 1.0, "tokens_per_sec": 4620, "mfu": 0.1229, "entropy": 0.720381511747837, "num_tokens": 23021901.0, "mean_token_accuracy": 0.8164495944976806, "epoch": 0.6935286935286935, "step": 1420}, {"loss": 0.6978864192962646, "grad_norm": 3.1219754219055176, "learning_rate": 2.990324263508726e-06, "padded_tokens": 23185652, "real_tokens": 23185652, "pad_ratio": 1.0, "tokens_per_sec": 4689, "mfu": 0.1248, "entropy": 0.6832813888788223, "num_tokens": 23185652.0, "mean_token_accuracy": 0.8254428178071975, "epoch": 0.6984126984126984, "step": 1430}, {"loss": 0.6963275909423828, "grad_norm": 3.046318531036377, "learning_rate": 2.9315551394142906e-06, "padded_tokens": 23345579, "real_tokens": 23345579, "pad_ratio": 1.0, "tokens_per_sec": 4622, "mfu": 0.123, "entropy": 0.7206508815288544, "num_tokens": 23345579.0, "mean_token_accuracy": 0.8214711725711823, "epoch": 0.7032967032967034, "step": 1440}, {"loss": 0.623183012008667, "grad_norm": 2.722724199295044, "learning_rate": 2.873428706659258e-06, "padded_tokens": 23506127, "real_tokens": 23506127, "pad_ratio": 1.0, "tokens_per_sec": 4643, "mfu": 0.1236, "entropy": 0.6125097997486592, "num_tokens": 23506127.0, "mean_token_accuracy": 0.8386768758296966, "epoch": 0.7081807081807082, "step": 1450}, {"loss": 0.6437493324279785, "grad_norm": 2.98553204536438, "learning_rate": 2.8159595099805937e-06, "padded_tokens": 23663242, "real_tokens": 23663242, "pad_ratio": 1.0, "tokens_per_sec": 4565, "mfu": 0.1215, "entropy": 0.7039372272789478, "num_tokens": 23663242.0, "mean_token_accuracy": 0.8284737408161164, "epoch": 0.7130647130647131, "step": 1460}, {"loss": 0.6209692478179931, "grad_norm": 2.562283754348755, "learning_rate": 2.7591619296578054e-06, "padded_tokens": 23823863, "real_tokens": 23823863, "pad_ratio": 1.0, "tokens_per_sec": 4692, "mfu": 0.1249, "entropy": 0.6635129131376744, "num_tokens": 23823863.0, "mean_token_accuracy": 0.8311059579253197, "epoch": 0.717948717948718, "step": 1470}, {"loss": 0.6717867374420166, "grad_norm": 2.788492202758789, "learning_rate": 2.7030501779146135e-06, "padded_tokens": 23987690, "real_tokens": 23987690, "pad_ratio": 1.0, "tokens_per_sec": 4722, "mfu": 0.1257, "entropy": 0.6884996920824051, "num_tokens": 23987690.0, "mean_token_accuracy": 0.8258467867970467, "epoch": 0.7228327228327228, "step": 1480}, {"loss": 0.6925924301147461, "grad_norm": 2.814424514770508, "learning_rate": 2.6476382953626855e-06, "padded_tokens": 24149879, "real_tokens": 24149879, "pad_ratio": 1.0, "tokens_per_sec": 4597, "mfu": 0.1223, "entropy": 0.6946474969387054, "num_tokens": 24149879.0, "mean_token_accuracy": 0.8241283476352692, "epoch": 0.7277167277167277, "step": 1490}, {"loss": 0.7131754875183105, "grad_norm": 2.451533555984497, "learning_rate": 2.5929401474883202e-06, "padded_tokens": 24313578, "real_tokens": 24313578, "pad_ratio": 1.0, "tokens_per_sec": 4687, "mfu": 0.1247, "entropy": 0.7106643125414849, "num_tokens": 24313578.0, "mean_token_accuracy": 0.8231417775154114, "epoch": 0.7326007326007326, "step": 1500}, {"eval_sft_loss": 0.6047070026397705, "eval_sft_runtime": 21.8689, "eval_sft_samples_per_second": 71.06, "eval_sft_steps_per_second": 8.917, "eval_entropy": 0.566621755201792, "eval_num_tokens": 24313578.0, "eval_mean_token_accuracy": 0.8460819490791596, "epoch": 0.7326007326007326, "step": 1500}, {"loss": 0.6297160625457764, "grad_norm": 2.4118845462799072, "learning_rate": 2.5389694211829347e-06, "padded_tokens": 24477304, "real_tokens": 24477304, "pad_ratio": 1.0, "tokens_per_sec": 12311, "mfu": 0.3276, "entropy": 0.6480682015419006, "num_tokens": 24477304.0, "mean_token_accuracy": 0.836219522356987, "epoch": 0.7374847374847375, "step": 1510}, {"loss": 0.642713212966919, "grad_norm": 7.135226726531982, "learning_rate": 2.485739621318255e-06, "padded_tokens": 24640999, "real_tokens": 24640999, "pad_ratio": 1.0, "tokens_per_sec": 4691, "mfu": 0.1248, "entropy": 0.6475187443196774, "num_tokens": 24640999.0, "mean_token_accuracy": 0.8329223081469536, "epoch": 0.7423687423687424, "step": 1520}, {"loss": 0.6718778133392334, "grad_norm": 3.523916006088257, "learning_rate": 2.433264067367047e-06, "padded_tokens": 24804721, "real_tokens": 24804721, "pad_ratio": 1.0, "tokens_per_sec": 4694, "mfu": 0.1249, "entropy": 0.6660301513969898, "num_tokens": 24804721.0, "mean_token_accuracy": 0.828752164542675, "epoch": 0.7472527472527473, "step": 1530}, {"loss": 0.7052314758300782, "grad_norm": 3.837459087371826, "learning_rate": 2.381555890070239e-06, "padded_tokens": 24966747, "real_tokens": 24966747, "pad_ratio": 1.0, "tokens_per_sec": 4667, "mfu": 0.1242, "entropy": 0.7083351090550423, "num_tokens": 24966747.0, "mean_token_accuracy": 0.8231446877121925, "epoch": 0.7521367521367521, "step": 1540}, {"loss": 0.6869406700134277, "grad_norm": 2.710775375366211, "learning_rate": 2.3306280281512727e-06, "padded_tokens": 25130442, "real_tokens": 25130442, "pad_ratio": 1.0, "tokens_per_sec": 4664, "mfu": 0.1241, "entropy": 0.686882919818163, "num_tokens": 25130442.0, "mean_token_accuracy": 0.8242148086428642, "epoch": 0.757020757020757, "step": 1550}, {"loss": 0.6714181900024414, "grad_norm": 3.5279605388641357, "learning_rate": 2.2804932250784983e-06, "padded_tokens": 25294115, "real_tokens": 25294115, "pad_ratio": 1.0, "tokens_per_sec": 4618, "mfu": 0.1229, "entropy": 0.6809089079499244, "num_tokens": 25294115.0, "mean_token_accuracy": 0.8289632499217987, "epoch": 0.7619047619047619, "step": 1560}, {"loss": 0.7630602836608886, "grad_norm": 3.337857961654663, "learning_rate": 2.231164025876427e-06, "padded_tokens": 25455430, "real_tokens": 25455430, "pad_ratio": 1.0, "tokens_per_sec": 4664, "mfu": 0.1241, "entropy": 0.7418067313730716, "num_tokens": 25455430.0, "mean_token_accuracy": 0.816575039923191, "epoch": 0.7667887667887668, "step": 1570}, {"loss": 0.68687744140625, "grad_norm": 2.996830463409424, "learning_rate": 2.18265277398665e-06, "padded_tokens": 25619107, "real_tokens": 25619107, "pad_ratio": 1.0, "tokens_per_sec": 4664, "mfu": 0.1241, "entropy": 0.6773345164954663, "num_tokens": 25619107.0, "mean_token_accuracy": 0.8266958922147751, "epoch": 0.7716727716727717, "step": 1580}, {"loss": 0.603700590133667, "grad_norm": 2.399977684020996, "learning_rate": 2.1349716081791773e-06, "padded_tokens": 25782876, "real_tokens": 25782876, "pad_ratio": 1.0, "tokens_per_sec": 4669, "mfu": 0.1243, "entropy": 0.6069677114486695, "num_tokens": 25782876.0, "mean_token_accuracy": 0.8435213372111321, "epoch": 0.7765567765567766, "step": 1590}]}