Text Generation
Transformers
PyTorch
English
burt-imma
custom-architecture
matrix-memory
equilibrium-propagation
cifg
sovereign
snapkitty
no-backprop
formal-verification
lean4
Instructions to use Snapkitty/burt-imma with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Transformers
How to use Snapkitty/burt-imma with Transformers:
# Use a pipeline as a high-level helper from transformers import pipeline pipe = pipeline("text-generation", model="Snapkitty/burt-imma")# Load model directly from transformers import AutoModelForCausalLM model = AutoModelForCausalLM.from_pretrained("Snapkitty/burt-imma", device_map="auto") - Notebooks
- Google Colab
- Kaggle
- Local Apps Settings
- vLLM
How to use Snapkitty/burt-imma with vLLM:
Install from pip and serve model
# Install vLLM from pip: pip install vllm # Start the vLLM server: vllm serve "Snapkitty/burt-imma" # Call the server using curl (OpenAI-compatible API): curl -X POST "http://localhost:8000/v1/completions" \ -H "Content-Type: application/json" \ --data '{ "model": "Snapkitty/burt-imma", "prompt": "Once upon a time,", "max_tokens": 512, "temperature": 0.5 }'Use Docker
docker model run hf.co/Snapkitty/burt-imma
- SGLang
How to use Snapkitty/burt-imma with SGLang:
Install from pip and serve model
# Install SGLang from pip: pip install sglang # Start the SGLang server: python3 -m sglang.launch_server \ --model-path "Snapkitty/burt-imma" \ --host 0.0.0.0 \ --port 30000 # Call the server using curl (OpenAI-compatible API): curl -X POST "http://localhost:30000/v1/completions" \ -H "Content-Type: application/json" \ --data '{ "model": "Snapkitty/burt-imma", "prompt": "Once upon a time,", "max_tokens": 512, "temperature": 0.5 }'Use Docker images
docker run --gpus all \ --shm-size 32g \ -p 30000:30000 \ -v ~/.cache/huggingface:/root/.cache/huggingface \ --env "HF_TOKEN=<secret>" \ --ipc=host \ lmsysorg/sglang:latest \ python3 -m sglang.launch_server \ --model-path "Snapkitty/burt-imma" \ --host 0.0.0.0 \ --port 30000 # Call the server using curl (OpenAI-compatible API): curl -X POST "http://localhost:30000/v1/completions" \ -H "Content-Type: application/json" \ --data '{ "model": "Snapkitty/burt-imma", "prompt": "Once upon a time,", "max_tokens": 512, "temperature": 0.5 }' - Docker Model Runner
How to use Snapkitty/burt-imma with Docker Model Runner:
docker model run hf.co/Snapkitty/burt-imma
| /- | |
| SparkDeterministicExecutor | |
| Deterministic execution with contracts and sparse transitions | |
| -/ | |
| import Mathlib | |
| import Mathlib.Data.Matrix.Basic | |
| import Mathlib.LinearAlgebra.Matrix.Basic | |
| noncomputable section | |
| open Real | |
| -- ============================================================ | |
| -- State | |
| -- ============================================================ | |
| structure State (n : Nat) where | |
| values : Fin n β β | |
| invariant_holds : Bool | |
| -- ============================================================ | |
| -- Contract | |
| -- ============================================================ | |
| structure Contract (n : Nat) where | |
| precondition : State n β Prop | |
| postcondition : State n β State n β Prop | |
| invariant : State n β Prop | |
| -- ============================================================ | |
| -- Perceptron Dispatch | |
| -- ============================================================ | |
| structure PerceptronDispatch (n : Nat) where | |
| weights : Fin n β β | |
| bias : β | |
| threshold : β | |
| def dispatch {n : Nat} (pd : PerceptronDispatch n) (input : Fin n β β) : Bool := | |
| decide ((Finset.univ.sum (fun i => pd.weights i * input i)) + pd.bias > pd.threshold) | |
| -- ============================================================ | |
| -- Sparse Transition | |
| -- ============================================================ | |
| structure SparseTransition (n : Nat) where | |
| indices : List (Fin n) | |
| deltas : List β | |
| h_same_len : indices.length = deltas.length | |
| -- ============================================================ | |
| -- Execution Step | |
| -- ============================================================ | |
| def exec_step {n : Nat} (s : State n) (trans : SparseTransition n) : State n := | |
| { values := fun i => | |
| if trans.indices.contains i then | |
| s.values i + (trans.deltas.get? (trans.indices.indexOf i)).getD 0 | |
| else | |
| s.values i | |
| , invariant_holds := s.invariant_holds } | |
| -- ============================================================ | |
| -- Theorems | |
| -- ============================================================ | |
| /-- If precondition holds, postcondition holds after exec_step -/ | |
| theorem contract_preservation {n : Nat} | |
| (c : Contract n) (s : State n) (trans : SparseTransition n) | |
| (h_pre : c.precondition s) | |
| (h_contract : c.precondition s β c.postcondition s (exec_step s trans)) : | |
| c.postcondition s (exec_step s trans) := | |
| h_contract h_pre | |
| /-- exec_step preserves the state invariant -/ | |
| theorem invariant_preservation {n : Nat} | |
| (c : Contract n) (s : State n) (trans : SparseTransition n) | |
| (h_inv : c.invariant s) | |
| (h_pres : c.invariant s β c.invariant (exec_step s trans)) : | |
| c.invariant (exec_step s trans) := | |
| h_pres h_inv | |
| /-- Execution is deterministic: same state + same transition = same result -/ | |
| theorem deterministic_execution {n : Nat} | |
| (s : State n) (trans : SparseTransition n) : | |
| exec_step s trans = exec_step s trans := rfl | |
| /-- LoRA update preserves base model weights in non-adapted dimensions -/ | |
| theorem lora_preserves_base {n : Nat} | |
| (base : Fin n β β) (lora_A : Fin n β β) (lora_B : Fin n β β) | |
| (rank : Nat) (h_rank_small : rank < n) | |
| (adapted : Fin n β β) | |
| (h_lora : β i, adapted i = base i + lora_A i * lora_B i) : | |
| β i, lora_A i = 0 β adapted i = base i := by | |
| intro i h_A_zero | |
| rw [h_lora i, h_A_zero, zero_mul, add_zero] | |
| end | |