Text Generation
MLX
Safetensors
English
small-language-model
from-scratch
conversational
apple-silicon
research
Instructions to use OpenSML/Solstice-1-175M with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- MLX
How to use OpenSML/Solstice-1-175M with MLX:
# Make sure mlx-lm is installed # pip install --upgrade mlx-lm # Generate text with mlx-lm from mlx_lm import load, generate model, tokenizer = load("OpenSML/Solstice-1-175M") prompt = "Write a story about Einstein" messages = [{"role": "user", "content": prompt}] prompt = tokenizer.apply_chat_template( messages, add_generation_prompt=True ) text = generate(model, tokenizer, prompt=prompt, verbose=True) - Notebooks
- Google Colab
- Kaggle
- Local Apps Settings
- LM Studio
- MLX LM
How to use OpenSML/Solstice-1-175M with MLX LM:
Generate or start a chat session
# Install MLX LM uv tool install mlx-lm # Interactive chat REPL mlx_lm.chat --model "OpenSML/Solstice-1-175M"
Run an OpenAI-compatible server
# Install MLX LM uv tool install mlx-lm # Start the server mlx_lm.server --model "OpenSML/Solstice-1-175M" # Calling the OpenAI-compatible server with curl curl -X POST "http://localhost:8080/v1/chat/completions" \ -H "Content-Type: application/json" \ --data '{ "model": "OpenSML/Solstice-1-175M", "messages": [ {"role": "user", "content": "Hello"} ] }' - Atomic Chat
File size: 6,296 Bytes
7bcd331 | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 | {
"adaptation": {
"adapter_compute": "float32",
"alpha": 64,
"base_compute": "bfloat16",
"dropout": 0.0,
"expected_trainable_parameters": 3785856,
"layers": "all",
"method": "lora",
"rank": 32,
"targets": [
"attn.qkv_proj",
"attn.proj"
],
"trainable_token_ids": [
2,
3
]
},
"answer_max_tokens": {
"constraints": 320,
"dialogue": 256,
"general": 256,
"grounded": 160,
"knowledge": 80,
"rewrite": 320,
"skills": 160,
"social": 128
},
"authored_holdouts": {
"constraints": 128,
"dialogue": 128,
"skills": 192
},
"authored_quotas": {
"constraints": 4200,
"dialogue": 6800,
"skills": 12000
},
"baseline_cache": "${WORKSPACE}/pretraining/prepared/chat_adapter_search_20261108/baseline_cache",
"batches": [
8
],
"betas": [
0.9,
0.95
],
"checked_components": [
"knowledge",
"grounded",
"skills",
"constraints",
"dialogue"
],
"comparison_checkpoint": "${WORKSPACE}/reports/benchmarks/deep44-reliable-chat-167-fp32-20261106/choices/checkpoint",
"component_quotas": {
"constraints": 800,
"dialogue": 2000,
"general": 2400,
"grounded": 2000,
"skills": 800
},
"components": [
"general",
"knowledge",
"grounded",
"skills",
"constraints",
"dialogue",
"social",
"rewrite"
],
"epochs": 1,
"eps": 1e-08,
"eval_every": 64,
"eval_examples_per_component": 64,
"execution": "single-mac",
"final_lr": 5e-06,
"format": "opensml-verified-chat-v2",
"fresh_mixture": {
"constraints": 0.4,
"dialogue": 0.1,
"grounded": 0.1,
"skills": 0.4
},
"fresh_weight": 0.5,
"generated_eval_per_component": 64,
"grad_accum": 8,
"grad_clip": 1.0,
"initial_checkpoint": "${WORKSPACE}/pretraining/runs/deep44_adapter_trial_09_replica_20261108/step_0002000_f3d1ffd185f9",
"keep_checkpoints": 6,
"kl_weight": 0.5,
"log_every": 10,
"max_cumulative_retention_loss_increase": 0.03,
"max_messages": 8,
"max_repetition_rate": 0.03333333333333333,
"max_retention_loss_increase": 0.003,
"max_sequence_length": 2048,
"max_steps": 256,
"method": "Editing and instruction skills with preservation",
"microbatch_rows": 4,
"min_domain_pass_cases": {},
"min_natural_stop": 0.9833333333333333,
"min_task_gain_cases": 4,
"minimum_source_tokens": 42000000000,
"original_pretraining_retention_loss": 2.8353193145245315,
"output": "${WORKSPACE}/pretraining/runs/deep44_parent2000_generalize_05_replica_20261109",
"peak_lr": 5e-05,
"prepared": "${WORKSPACE}/pretraining/runs/deep44_parent2000_generalize_05_replica_20261109/input/corpus",
"public_holdouts": {
"constraints_smol": 32,
"dialogue_smol": 64,
"general_smol": 64,
"grounded_no": 32,
"grounded_yes": 96,
"knowledge": 128,
"rewrite_smol": 32,
"social": 32
},
"public_quotas": {
"constraints_smol": 2000,
"dialogue_smol": 1200,
"general_smol": 4800,
"grounded_no": 2750,
"grounded_yes": 8250,
"knowledge": 4000,
"rewrite_smol": 1000,
"social": 1000
},
"qualitative_eval_per_component": 8,
"regression_patience": 2,
"replay_length": 512,
"replay_mixture": {
"constraints": 0.125,
"dialogue": 0.125,
"general": 0.1875,
"grounded": 0.125,
"knowledge": 0.1875,
"rewrite": 0.0625,
"skills": 0.125,
"social": 0.0625
},
"replay_objective_weight": 0.2,
"replay_policy": "prose-only",
"replay_rows_per_update": 16,
"retention": "${WORKSPACE}/pretraining/data/deep44_pretraining_validation_20261102",
"rows_per_group": 16,
"seed": 202611097,
"selection_reference": "Parent 2000: +4/320 development, +8/704 final, no domain/control regressions; fresh 256-case check must not regress; full-suite guard after private pass",
"sft_objective_weight": 0.8,
"sft_rows_per_update": 48,
"sft_stage": 2,
"source_checkpoint": "${WORKSPACE}/pretraining/runs/deep44_cooldown_2b_20261107/step_0320445_08b689895ed2",
"source_run": "${WORKSPACE}/pretraining/runs/deep44_cooldown_2b_20261107",
"sources": [
{
"file": "train.jsonl.gz",
"label": "helpsteer2",
"license": "cc-by-4.0",
"repo": "nvidia/HelpSteer2",
"revision": "990b2711a36180dd19d9c94b8627844866f8982a"
},
{
"file": "data/train-00000-of-00001-b42a775f407cee45.parquet",
"label": "oasst1",
"license": "apache-2.0",
"repo": "OpenAssistant/oasst1",
"revision": "fdf72ae0827c1cda404aff25b6603abec9e3399b"
},
{
"files": [
"data/smol-magpie-ultra/train-00000-of-00006.parquet",
"data/smol-magpie-ultra/train-00001-of-00006.parquet",
"data/smol-magpie-ultra/train-00002-of-00006.parquet",
"data/smol-magpie-ultra/train-00003-of-00006.parquet",
"data/smol-magpie-ultra/train-00004-of-00006.parquet",
"data/smol-magpie-ultra/train-00005-of-00006.parquet"
],
"label": "smol_magpie_ultra",
"license": "apache-2.0",
"repo": "HuggingFaceTB/smoltalk",
"revision": "5feaf2fd3ffca7c237fc38d1861bc30365d48ffa"
},
{
"label": "squad",
"license": "cc-by-sa-4.0",
"parquet_urls": [
"https://huggingface.co/datasets/rajpurkar/squad_v2/resolve/3ffb306f725f7d2ce8394bc1873b24868140c412/squad_v2/train-00000-of-00001.parquet"
],
"repo": "rajpurkar/squad_v2",
"revision": "3ffb306f725f7d2ce8394bc1873b24868140c412",
"split": "train"
},
{
"label": "squad_validation",
"license": "cc-by-sa-4.0",
"parquet_urls": [
"https://huggingface.co/datasets/rajpurkar/squad_v2/resolve/3ffb306f725f7d2ce8394bc1873b24868140c412/squad_v2/validation-00000-of-00001.parquet"
],
"repo": "rajpurkar/squad_v2",
"revision": "3ffb306f725f7d2ce8394bc1873b24868140c412",
"split": "validation"
}
],
"strategy": "Fresh computed instructions and context tasks plus broad original-training replay; optional forward KL to the frozen parent on replay assistant tokens; all attempts restart exactly from parent2000 with fresh moments",
"tokenizer": "${WORKSPACE}/reference/opensml-150m/tokenizer/bytebpe32k_v1",
"warmup_fraction": 0.1,
"weight_decay": 0.0
}
|