Model reference · open weights
L3-Stheno-32K is an open-weight language model from Sao10K. AxForge deploys and operates it for you on dedicated EU-owned hardware — with the licence handled where one is required.
Available as managed deployment — configured and operated for you on dedicated EU hardware, quoted per deployment.
What it is
| Released by | Sao10K |
|---|---|
| Type | Language models |
| Task | Text gen |
| Parameters (lead) | 8.0B |
| Context | 32k tokens |
| Runs with | transformers |
| Released | 2024-06-22 |
| Popularity | 581 downloads / month |
| Licence | Commercial licence needed |
About
Trained with compute from Backyard.ai | Thanks to them and @dynafire for helping me out.
Training Details:
Dataset Modifications:
Notes:
Sanity Check // Needle in a Haystack Results:
Wandb Run:
Relevant Axolotl Configurations:
sequence_len: 8192
use_pose: true
pose_max_context_len: 32768
overrides_of_model_config:
rope_theta: 2000000.0
max_position_embeddings: 32768
# peft_use_dora: true
adapter: lora
peft_use_rslora: true
lora_model_dir:
lora_r: 256
lora_alpha: 256
lora_dropout: 0.1
lora_target_linear: true
lora_target_modules:
- gate_proj
- down_proj
- up_proj
- q_proj
- v_proj
- k_proj
- o_proj
warmup_steps: 80
gradient_accumulation_steps: 6
micro_batch_size: 1
num_epochs: 2
optimizer: adamw_bnb_8bit
lr_scheduler: cosine_with_min_lr
learning_rate: 0.00004
lr_scheduler_kwargs:
min_lr: 0.000004
From the published model card. Full card on the HuggingFace links in the sidebar.
Using it via the API
Once AxForge deploys l3-stheno-32k for you, it answers on the OpenAI-compatible API — the same base URL and keys as every other model. (l3-stheno-32k below is illustrative; you get the exact model name on deployment.)
$ curl -sS https://api.axforge.ai/v1/chat/completions \
-H "Authorization: Bearer $AXFORGE_API_KEY" \
-H "Content-Type: application/json" \
-d '{"model":"l3-stheno-32k","messages":[{"role":"user","content":"Hello"}]}'
Create an account — your API key is available in the console. 3M free tokens every 30 days with every new account.