wehub-resource-sync
caf324b09d
tests / check_code_quality (push) Waiting to run
tests / tests (ubuntu-latest, 3.10) (push) Blocked by required conditions
tests / tests (ubuntu-latest, 3.11) (push) Blocked by required conditions
Deploy "method_comparison" Gradio to Spaces / deploy (push) Waiting to run
Deploy "PEFT shop" Gradio app to Spaces / deploy (push) Waiting to run
tests on transformers main / tests (push) Waiting to run
tests / tests (ubuntu-latest, 3.12) (push) Blocked by required conditions
tests / tests (ubuntu-latest, 3.13) (push) Blocked by required conditions
tests / tests (windows-latest, 3.10) (push) Blocked by required conditions
tests / tests (windows-latest, 3.11) (push) Blocked by required conditions
tests / tests (windows-latest, 3.12) (push) Blocked by required conditions
tests / tests (windows-latest, 3.13) (push) Blocked by required conditions
Secret Leaks / trufflehog (push) Waiting to run
CI security linting / zizmor latest via Cargo (push) Waiting to run
Build documentation / build (push) Failing after 0s
29 lines
647 B
JSON
29 lines
647 B
JSON
{
|
|
"model_id": "meta-llama/Llama-3.2-3B",
|
|
"dtype": "bfloat16",
|
|
"max_seq_length": 768,
|
|
"batch_size": 4,
|
|
"batch_size_eval": 50,
|
|
"max_steps": 5000,
|
|
"eval_steps": 250,
|
|
"compile": false,
|
|
"use_gc": false,
|
|
"seed": 0,
|
|
"grad_norm_clip": 1.0,
|
|
"optimizer_type": "AdamW",
|
|
"optimizer_kwargs": {
|
|
"lr": 1e-4,
|
|
"weight_decay": 0.1
|
|
},
|
|
"lr_scheduler": "cosine",
|
|
"use_amp": false,
|
|
"autocast_adapter_dtype": true,
|
|
"attn_implementation": null,
|
|
"generation_kwargs": {
|
|
"max_length": 800,
|
|
"max_new_tokens": 300
|
|
},
|
|
"query_template": "Question: {query} Think step by step.\nAnswer:",
|
|
"init_kv_cache_prefix": null
|
|
}
|