ZeroAgency/zero-llama-3.1-8b-beta6

See axolotl config

axolotl version: 0.9.0

# zero-llama-3.1-8b-beta6
# [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
# [2025-05-05 20:54:51,708] [DEBUG] [axolotl.calculate_total_num_steps:423] [PID:15283] [RANK:0] `total_supervised_tokens: 868_324_310`
# adapter: lora
base_model: nvidia/Llama-3.1-Nemotron-8B-UltraLong-1M-Instruct
dataset_processes: 64
chat_template: jinja
chat_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"


dataset_prepared_path: ./last_run_prepared

datasets:
- message_property_mappings:
    content: content
    role: role
  path: ZeroAgency/ru-big-russian-dataset
  trust_remote_code: false
  field_messages: conversation
  type: chat_template

- message_property_mappings:
    content: content
    role: role
  path: bethrezen/shkolkovo-2
  trust_remote_code: false
  field_messages: messages
  type: chat_template

- message_property_mappings:
    content: content
    role: role
  path: bethrezen/thinking-summary-v2
  trust_remote_code: false
  field_messages: conversation
  type: chat_template

test_datasets:
- message_property_mappings:
    content: content
    role: role
  path: ZeroAgency/ru-big-russian-dataset
  trust_remote_code: false
  field_messages: conversation
  type: chat_template
  split: test

# approx 20k samples should be enough
#val_set_size: 0.061

# exact duplicates are already cleaned
#dataset_exact_deduplication: true

gradient_accumulation_steps: 1
#gradient_checkpointing: true
gradient_checkpointing_kwargs:
  use_reentrant: false

#learning_rate: 0.0001
learning_rate: 1e-5
lisa_layers_attribute: model.layers
#is_mistral_derived_model: true

plugins:
  - axolotl.integrations.liger.LigerPlugin
liger_rope: true
liger_rms_norm: true
liger_swiglu: true
liger_fused_linear_cross_entropy: true

# load_best_model_at_end: true
load_in_4bit: false
load_in_8bit: false
# lora_alpha: 256
# lora_dropout: 0
# lora_target_linear: true
# lora_r: 256

lr_scheduler: cosine
#max_prompt_len: 8192
mean_resizing_embeddings: false

micro_batch_size: 1
num_epochs: 3
optimizer: adamw_torch_fused
output_dir: ./outputs/zero-llama-3.1-8b-beta5


sample_packing_bin_size: 200
sample_packing_group_size: 100000
save_only_model: false
save_safetensors: true
sequence_len: 8196
min_sample_len: 1
shuffle_merged_datasets: true
skip_prepare_dataset: false
strict: false
train_on_inputs: false


weight_decay: 0.01
wandb_project: Zero-Llama
wandb_name: zero-llama-3.1-8b-beta5
bf16: true
fp16: false
tf32: false
flash_attention: true

save_strategy: epoch
eval_strategry: epoch

logging_steps: 1
save_total_limit: 2
warmup_steps: 100
sample_packing: true
pad_to_sequence_len: true
group_by_length: true
seed: 42
data_seed: 42

deepspeed: /workspace/axolotl/deepspeed_configs/zero1_torch_compile.json
log_with: wandb
trust_remote_code: true
use_fast_tokenizer: true
special_tokens:
  pad_token: <|finetune_right_pad_id|>
  eos_token: <|eot_id|>
#torch_compile: auto

outputs/zero-llama-3.1-8b-beta5

This model is a fine-tuned version of nvidia/Llama-3.1-Nemotron-8B-UltraLong-1M-Instruct on the ZeroAgency/ru-big-russian-dataset, the bethrezen/shkolkovo-2, and the bethrezen/thinking-summary-v2 datasets. It achieves the following results on the evaluation set:

• Loss: 1.0706

Model description

More information needed

Intended uses & limitations

More information needed

Training and evaluation data

More information needed

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

• learning_rate: 1e-05
• train_batch_size: 1
• eval_batch_size: 1
• seed: 42
• distributed_type: multi-GPU
• num_devices: 8
• total_train_batch_size: 8
• total_eval_batch_size: 8
• optimizer: Use OptimizerNames.ADAMW_TORCH_FUSED with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: cosine
• lr_scheduler_warmup_steps: 100
• num_epochs: 3.0

Training results

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

Framework versions

• Transformers 4.51.3
• Pytorch 2.6.0+cu124
• Datasets 3.5.0
• Tokenizers 0.21.1