| Filename | Latest commit message | Latest commit date |
|---|---|---|
a6ee726b2f
The fabric arc reaches the standard's receipts. The hybrid
backend wraps every complete() in an attribution-ledger
window: GFLOPs recorded per fabric member that ACTUALLY ran
each matmul, priced by the machine's IOReport calibration.
The summed figure rides LlmResponse.energy_joules, and the
L3 stage now grades it correctly per SPEC §6:
backend reports a figure → estimated_uj, ModelBased
(a calibrated per-operation
model — precisely what §6
defines ModelBased to mean)
no figure → catalog constant, Estimator
(the honest old behavior,
unchanged for Echo/llama.cpp)
Live, real checkpoint (Mellum2, fabric ON):
"The capital of France is" → " Paris. \nParis is"
fabric-priced energy: 113.0 J (ModelBased)
A coding-cascade L3 receipt now carries energy from the
silicon that actually executed the call — attributed per
member, priced by a calibration metered on this machine —
instead of a 1.0 J placeholder. The provenance ladder is
exercised end to end: Estimator (catalog) → ModelBased
(fabric ledger × calibration) → HwShunt (a meter wrapping
the whole call, as the joule-fix harness already does).
Test updated to the new contract: Some(>0) with the fabric
on, honest None without — both branches printed.
Co-Authored-By: Claude Fable 5 <[email protected]>
|
||
| .. | ||
| src | ||
| tests | ||
| Cargo.lock | ||
| Cargo.toml | ||
| README.md | ||
L3 binding — JouleClaw model tier → Joule Code cascade
A consumer binding example: how to wire JouleClaw's L3 (Model) tier into the Joule Code cascade. It is a standalone workspace, so the standard's reference library keeps no JouleClaw dependency.
A coding request that misses the cheaper tiers (L0 cache, L1 lawful, L2 embed) is
generated by a local model at L3, producing a content-addressed CodingOp and a
joules-class Receipt.
Three backends, same slot
All implement JouleClaw's jouleclaw_llm_cheap::LlmBackend, so they drop into
the same cascade stage:
EchoBackend(from JouleClaw) — deterministic, model-free; used by the unit tests so the wiring is verifiable with no model present.JouleclawRuntimeBackend(here) — runs the model in-process on JouleClaw's own from-scratch GGUF tensor runtime (jouleclaw-runtime): no external binary, no FFI. Tokenize → build the decode-step graph → compile against the registered kernels (jouleclaw-backend-reference/-apple) → autoregressive decode → detokenize, all in Rust the family wrote. This is the default backend for therunbinary.LlamaCompletionBackend(here) — drives a real on-device GGUF viallama-completion(llama.cpp). SetJOULE_CODE_L3_ENGINE=llamato use it.
Either local path: no datacenter, no network. In production the cheapest capable
backend is chosen by jouleclaw_dispatch::ModelBank's cost table.
Run it against a local model
# Default: in-process on jouleclaw-runtime (the from-scratch tensor engine).
cargo run --release --bin run -- /path/to/model.gguf "Write a Rust function that reverses a string."
# or: JOULE_CODE_L3_MODEL=/path/to/model.gguf cargo run --release --bin run
# Use the llama.cpp shell-out backend instead:
JOULE_CODE_L3_ENGINE=llama cargo run --release --bin run -- /path/to/model.gguf "..."
The run binary generates with the local model, content-addresses the result as
a CodingOp (tier L3), and meters host energy over the inference with
JouleClaw's energy counter — so the receipt's joule cost is a real local
measurement, honestly tagged with the counter's provenance (HwShunt on
RAPL/NVML hosts; Estimator on Apple Silicon until IOReport is wired), not a
datacenter token estimate.
Verified live: the llama.cpp path against a local 350M GGUF, and the in-process
from-scratch runtime against ternary-bonsai-1.7b-q2_0 (qwen3, Q2_0) — real text
generated, a real op_id, the cascade closing at L3.
Tests
cargo test # EchoBackend path — no model required
# In-process from-scratch runtime against a real GGUF (opt-in):
JOULE_BONSAI=/path/to/ternary-bonsai-1.7b-q2_0.gguf \
cargo test --release --test runtime_l3 -- --ignored --nocapture