Views
No views yet
0.9.01# zero-llama-3.1-8b-beta6
2# [2025-05-05 20:54:31,355] [DEBUG] [axolotl.calculate_total_num_steps:405] [PID:15283] [RANK:0] total_num_tokens: 1_374_129_930
3# [2025-05-05 20:54:51,708] [DEBUG] [axolotl.calculate_total_num_steps:423] [PID:15283] [RANK:0] `total_supervised_tokens: 868_324_310`
4# adapter: lora
5base_model: nvidia/Llama-3.1-Nemotron-8B-UltraLong-1M-Instruct
6dataset_processes: 64
7chat_template: jinja
8chat_template_jinja: "{{- bos_token }}\n{%- if custom_tools is defined %}\n {%- set tools = custom_tools %}\n{%- endif %}\n{%- if not tools_in_user_message is defined %}\n {%- set tools_in_user_message = true %}\n{%- endif %}\n{%- if not date_string is defined %}\n {%- set date_string = \"26 Jul 2024\" %}\n{%- endif %}\n{%- if not tools is defined %}\n {%- set tools = none %}\n{%- endif %}\n\n{#- This block extracts the system message, so we can slot it into the right place. #}\n{%- if messages[0]['role'] == 'system' %}\n {%- set system_message = messages[0]['content']|trim %}\n {%- set messages = messages[1:] %}\n{%- else %}\n {%- set system_message = \"\" %}\n{%- endif %}\n\n{#- System message + builtin tools #}\n{{- \"<|start_header_id|>system<|end_header_id|>\\n\\n\" }}\n{%- if builtin_tools is defined or tools is not none %}\n {{- \"Environment: ipython\\n\" }}\n{%- endif %}\n{%- if builtin_tools is defined %}\n {{- \"Tools: \" + builtin_tools | reject('equalto', 'code_interpreter') | join(\", \") + \"\\n\\n\"}}\n{%- endif %}\n{{- \"Cutting Knowledge Date: December 2023\\n\" }}\n{{- \"Today Date: \" + date_string + \"\\n\\n\" }}\n{%- if tools is not none and not tools_in_user_message %}\n {{- \"You have access to the following functions. To call a function, please respond with JSON for a function call.\" }}\n {{- 'Respond in the format {\"name\": function name, \"parameters\": dictionary of argument name and its value}.' }}\n {{- \"Do not use variables.\\n\\n\" }}\n {%- for t in tools %}\n {{- t | tojson(indent=4) }}\n {{- \"\\n\\n\" }}\n {%- endfor %}\n{%- endif %}\n{{- system_message }}\n{{- \"<|eot_id|>\" }}\n\n{#- Custom tools are passed in a user message with some extra guidance #}\n{%- if tools_in_user_message and not tools is none %}\n {#- Extract the first user message so we can plug it in here #}\n {%- if messages | length != 0 %}\n {%- set first_user_message = messages[0]['content']|trim %}\n {%- set messages = messages[1:] %}\n {%- else %}\n {{- raise_exception(\"Cannot put tools in the first user message when there's no first user message!\") }}\n{%- endif %}\n {{- '<|start_header_id|>user<|end_header_id|>\\n\\n' -}}\n {{- \"Given the following functions, please respond with a JSON for a function call \" }}\n {{- \"with its proper arguments that best answers the given prompt.\\n\\n\" }}\n {{- 'Respond in the format {\"name\": function name, \"parameters\": dictionary of argument name and its value}.' }}\n {{- \"Do not use variables.\\n\\n\" }}\n {%- for t in tools %}\n {{- t | tojson(indent=4) }}\n {{- \"\\n\\n\" }}\n {%- endfor %}\n {{- first_user_message + \"<|eot_id|>\"}}\n{%- endif %}\n\n{%- for message in messages %}\n {%- if not (message.role == 'ipython' or message.role == 'tool' or 'tool_calls' in message) %}\n {{- '<|start_header_id|>' + message['role'] + '<|end_header_id|>\\n\\n'+ message['content'] | trim + '<|eot_id|>' }}\n {%- elif 'tool_calls' in message %}\n {%- if not message.tool_calls|length == 1 %}\n {{- raise_exception(\"This model only supports single tool-calls at once!\") }}\n {%- endif %}\n {%- set tool_call = message.tool_calls[0].function %}\n {%- if builtin_tools is defined and tool_call.name in builtin_tools %}\n {{- '<|start_header_id|>assistant<|end_header_id|>\\n\\n' -}}\n {{- \"<|python_tag|>\" + tool_call.name + \".call(\" }}\n {%- for arg_name, arg_val in tool_call.arguments | items %}\n {{- arg_name + '=\"' + arg_val + '\"' }}\n {%- if not loop.last %}\n {{- \", \" }}\n {%- endif %}\n {%- endfor %}\n {{- \")\" }}\n {%- else %}\n {{- '<|start_header_id|>assistant<|end_header_id|>\\n\\n' -}}\n {{- '{\"name\": \"' + tool_call.name + '\", ' }}\n {{- '\"parameters\": ' }}\n {{- tool_call.arguments | tojson }}\n {{- \"}\" }}\n {%- endif %}\n {%- if builtin_tools is defined %}\n {#- This means we're in ipython mode #}\n {{- \"<|eom_id|>\" }}\n {%- else %}\n {{- \"<|eot_id|>\" }}\n {%- endif %}\n {%- elif message.role == \"tool\" or message.role == \"ipython\" %}\n {{- \"<|start_header_id|>ipython<|end_header_id|>\\n\\n\" }}\n {%- if message.content is mapping or message.content is iterable %}\n {{- message.content | tojson }}\n {%- else %}\n {{- message.content }}\n {%- endif %}\n {{- \"<|eot_id|>\" }}\n {%- endif %}\n{%- endfor %}\n{%- if add_generation_prompt %}\n {{- '<|start_header_id|>assistant<|end_header_id|>\\n\\n' }}\n{%- endif %}\n"
9
10
11dataset_prepared_path: ./last_run_prepared
12
13datasets:
14- message_property_mappings:
15 content: content
16 role: role
17 path: ZeroAgency/ru-big-russian-dataset
18 trust_remote_code: false
19 field_messages: conversation
20 type: chat_template
21
22- message_property_mappings:
23 content: content
24 role: role
25 path: bethrezen/shkolkovo-2
26 trust_remote_code: false
27 field_messages: messages
28 type: chat_template
29
30- message_property_mappings:
31 content: content
32 role: role
33 path: bethrezen/thinking-summary-v2
34 trust_remote_code: false
35 field_messages: conversation
36 type: chat_template
37
38test_datasets:
39- message_property_mappings:
40 content: content
41 role: role
42 path: ZeroAgency/ru-big-russian-dataset
43 trust_remote_code: false
44 field_messages: conversation
45 type: chat_template
46 split: test
47
48# approx 20k samples should be enough
49#val_set_size: 0.061
50
51# exact duplicates are already cleaned
52#dataset_exact_deduplication: true
53
54gradient_accumulation_steps: 1
55#gradient_checkpointing: true
56gradient_checkpointing_kwargs:
57 use_reentrant: false
58
59#learning_rate: 0.0001
60learning_rate: 1e-5
61lisa_layers_attribute: model.layers
62#is_mistral_derived_model: true
63
64plugins:
65 - axolotl.integrations.liger.LigerPlugin
66liger_rope: true
67liger_rms_norm: true
68liger_swiglu: true
69liger_fused_linear_cross_entropy: true
70
71# load_best_model_at_end: true
72load_in_4bit: false
73load_in_8bit: false
74# lora_alpha: 256
75# lora_dropout: 0
76# lora_target_linear: true
77# lora_r: 256
78
79lr_scheduler: cosine
80#max_prompt_len: 8192
81mean_resizing_embeddings: false
82
83micro_batch_size: 1
84num_epochs: 3
85optimizer: adamw_torch_fused
86output_dir: ./outputs/zero-llama-3.1-8b-beta5
87
88
89sample_packing_bin_size: 200
90sample_packing_group_size: 100000
91save_only_model: false
92save_safetensors: true
93sequence_len: 8196
94min_sample_len: 1
95shuffle_merged_datasets: true
96skip_prepare_dataset: false
97strict: false
98train_on_inputs: false
99
100
101weight_decay: 0.01
102wandb_project: Zero-Llama
103wandb_name: zero-llama-3.1-8b-beta5
104bf16: true
105fp16: false
106tf32: false
107flash_attention: true
108
109save_strategy: epoch
110eval_strategry: epoch
111
112logging_steps: 1
113save_total_limit: 2
114warmup_steps: 100
115sample_packing: true
116pad_to_sequence_len: true
117group_by_length: true
118seed: 42
119data_seed: 42
120
121deepspeed: /workspace/axolotl/deepspeed_configs/zero1_torch_compile.json
122log_with: wandb
123trust_remote_code: true
124use_fast_tokenizer: true
125special_tokens:
126 pad_token: <|finetune_right_pad_id|>
127 eos_token: <|eot_id|>
128#torch_compile: auto
129
130| Training Loss | Epoch | Step | Validation Loss |
|---|---|---|---|
| 0.6077 | 1.0 | 22970 | 0.9736 |
| 0.3958 | 2.0 | 45940 | 0.9737 |
| 0.3008 | 3.0 | 68910 | 1.0706 |