saklas

Saklas is a local workbench for mechanistic interpretability on large language models.

It comes with a local dashboard, along with a Python API, and a server compatible with both OpenAI and Ollama.

Quick start

pip install saklas
saklas serve google/gemma-3-4b-it

Open http://localhost:8000.

The first launch downloads the model and fits the 17 bundled concept probes. This can take a while; they get stored in ~/.saklas/ for future launches.

For NVIDIA CUDA:

pip install "saklas[cuda,flash]"
saklas serve google/gemma-3-4b-it --device cuda

WebUI

Threads

Every authored or generated turn becomes a node in a branching tree. You can reroll from any point and the tree lets you save and load conversations.

The bottom of the panel has the standard sampling controls: temperature, top-p, top-k, repetition and presence penalties, and everything else you'd expect.

Completion modes

Chat mode renders the model's template as turns with collapsible thinking. Roles are handled dynamically:

• you write selects the role the text you write will be appended as;
• model writes selects the role the model continues as (or none to just append);
• the cast button lets you assign steering to specific roles.

Compatible chat templates support arbitrary role labels beyond user and assistant.

Raw mode exposes one raw buffer for base models.

Token-level inspection

Tokens can be highlighted by probe scores or surprise (logprob).

Clicking on a token opens one detail drawer with four views:

• geometry — attached probe readings across layers;
• logits — the chosen token and captured alternatives with logprobs, plus token forking;
• SAE — sparse feature activations;
• J-lens — the aggregate workspace and layer-by-vocabulary readout.

The detail cursor can walk tokens, thinking/response segments, and turns without closing the drawer. Historical rows use their captured measurements when present and can replay the producing prefix for newly attached instrumentation.

Clicking on the info icon by a probe shows a fitted concept's geometry layer by layer.

Instruments

The right side of the workbench has four tabs.

Analysis tools

Press Cmd+K or Ctrl+K elsewhere to open a menu with further options:

• build, fit, merge, install, and inspect manifolds;
• author and score restricted-choice templates;
• manage the cast and open steering workflows;
• inspect correlations and pairwise layer geometry;
• check model, device, source, authentication, and server health;
• open the built-in help surface.

Concepts, subspaces, and manifolds

In Saklas, you extract concepts as manifolds or subspaces.

• A 1D flat subspace is just a steering vector.
• A higher-rank flat subspace is a group of orthogonal steering vectors.
• A curved manifold fits a nonlinear surface and lets you steer along it.

Bundled concepts

Saklas comes with 17 concept pairs that are attached as probes by default:

Three larger concept sets ship as well, but aren't fitted by default:

• personas — 107 personas;
• emotions — 20 emotional states;
• months — 12 months.

Steering expressions

Every surface uses the same expression format. The same recipe can be used across Python, YAML, OpenAI, Ollama, or the native API.

0.3 honest + 0.4 warm
0.5 formal - 0.2 verbose
0.3 honest|sycophantic
0.3 honest~confident
!sycophantic
0.5 personas%pirate
0.7,0.4 months%january@response
0.4 warm@when:confident.uncertain>0.4
0.3 jlens/orange + 0.2 sae/9143

Manifold coefficients use two coordinates: along and onto. along controls movement within the manifold toward the target; onto reduces the off-surface component inside the manifold's fitted tube.

How Saklas works

Extraction

Saklas first has the model answer a shared set of baseline prompts as each concept, then it takes the resulting hidden states and fits them to either a curved manifold or a flat subspace.

Layer allocation uses a Mahalanobis metric estimated from neutral activations. Discriminative layer selection removes flat axes that fail to straddle the neutral baseline across the fitted nodes.

The full data flow, artifact boundaries, instrument protocol, and concurrency invariants are documented in ARCHITECTURE.md.

Monitoring

A reading includes fitted coordinates, the centered activation's subspace fraction, the nearest nodes and their soft assignment, and, for curved manifolds, an off-surface residual and tube membership.

Jacobian lens

The Jacobian lens implementation follows Gurnee et al.'s work. Saklas lets you use one of the published J-lens artifacts, or fit your own.

Sparse autoencoders

Saklas can either use a published SAELens release or train a local SAE.

Installation

Saklas requires Python 3.11 or newer and PyTorch 2.2 or newer. CUDA or Apple Silicon MPS is strongly recommended for interactive use; CPU is supported for smaller models and non-GPU workflows.

pip install saklas

The base package includes the HTTP server, the prebuilt Svelte WebUI, and SAELens. Optional extras add specialized workflows:

Extras can be combined:

pip install "saklas[cuda,flash]"       # full tested NVIDIA path
pip install "saklas[hf,research]"      # dataset-backed research workflows
pip install "saklas[notebook]"         # interactive figures

cuda and flash are Linux/NVIDIA CUDA extras. FlashAttention is selected automatically when installed; there is no runtime flag to enable it.

From source:

git clone https://github.com/a9lim/saklas
cd saklas
pip install -e ".[dev]"

Running the server

saklas serve MODEL [options]

Common options:

serve and every subcommand that accepts -c/--config read ~/.saklas/config.yaml first and then compose any explicit -c PATH files on top. For example:

model: google/gemma-3-4b-it
vectors: "0.3 honest + 0.2 warm"
temperature: 0.8
top_p: 0.9
max_tokens: 512
return_top_k: 8

Inspect the resolved configuration with saklas config show and validate a file with saklas config validate path.yaml.

Command-line artifact workflows

The CLI has eight top-level verbs. The WebUI covers the interactive versions of most workflows; the CLI is useful for reproducible preparation, distribution, and batch work.

Representative commands:

# Extract and fit a two-pole concept for one model
saklas manifold extract patient impatient -m google/gemma-3-4b-it

# Fit a bundled many-node manifold
saklas manifold fit personas -m google/gemma-3-4b-it

# Install or publish manifold packs through Hugging Face
saklas pack search creativity
saklas pack install OWNER/REPO
saklas pack push local/patient.impatient -a OWNER/REPO -m google/gemma-3-4b-it

# Fetch a provider J-LENS or fit a local R-lens (RelP is the fit default)
saklas lens fetch google/gemma-3-4b-it
saklas lens fit org/model --prompts 100
# Use --standard when you specifically want local:default instead

# Fetch an SAE or train a local source
saklas sae fetch google/gemma-3-4b-it saelens:gemma-scope-2-4b-it-res
saklas sae train org/model my-sae --layer 20 --tokens 1000000

Run saklas <verb> -h and saklas <verb> <subcommand> -h for the complete flag surface.

HTTP APIs

The same saklas serve process exposes four surfaces on one port:

• / — the Saklas WebUI;
• /v1/* — OpenAI-compatible models and chat completions;
• /api/* — Ollama-compatible generation and chat;
• /saklas/v1/* — native sessions, loom trees, probes, manifolds, templates, SAE/J-LENS lifecycle and replay, SSE, and token-plus-measurement WebSockets.

Interactive OpenAPI documentation is available at http://localhost:8000/docs.

OpenAI client

from openai import OpenAI

client = OpenAI(base_url="http://localhost:8000/v1", api_key="unused")
response = client.chat.completions.create(
    model="google/gemma-3-4b-it",
    messages=[{"role": "user", "content": "Describe a rainy afternoon."}],
    extra_body={"steering": "0.3 warm + 0.2 concise"},
)
print(response.choices[0].message.content)

Ollama client

curl -N http://localhost:8000/api/chat -d '{
  "model": "gemma3",
  "messages": [{"role": "user", "content": "Write a short haiku."}],
  "options": {"steer": "0.3 warm - 0.2 formal.casual"}
}'

Saklas targets a trusted local machine or lab network. It is not a hardened multi-tenant inference service. If you bind it beyond a trusted host, read SECURITY.md, set an API key, and add TLS, rate limits, request limits, and isolation outside Saklas.

Python API

from saklas import SaklasSession, SamplingConfig

with SaklasSession.from_pretrained(
    "google/gemma-3-4b-it",
    device="auto",
    return_top_k=8,
) as session:
    name, profile = session.extract("patient", baseline="impatient")
    session.add_probe(name)

    result = session.generate(
        "How should I learn a difficult skill?",
        steering=f"0.3 {name} + 0.2 concise",
        sampling=SamplingConfig(
            temperature=0.8,
            top_p=0.9,
            max_tokens=256,
            seed=42,
        ),
    ).first

    print(result.text)
    print(result.applied_steering)
    print(result.probe_readings[name].coords)

generate and generate_stream accept the same steering expression as the WebUI. Generation returns a list-like RunSet; .first is convenient for a single completion. GenerationResult carries text, token IDs, throughput and timing, finish reason, the canonical applied expression, captured log probabilities, aggregate probe readings, and the versioned measurement envelope every instrument family writes into.

Batch and sweep helpers return the same result shape:

batch = session.generate_batch(
    ["Describe a sunset.", "Describe a storm."],
    steering="0.3 warm",
)

sweep = session.generate_sweep(
    "Describe a forest.",
    sweep={"warm.clinical": [-0.4, 0.0, 0.4]},
)

Restricted-choice scoring evaluates a candidate distribution directly under the model, optionally with steering:

scores = session.score_choices(
    [{"role": "user", "content": "The first weekday is"}],
    ["Monday", "Tuesday", "Wednesday"],
    steering="0.3 confident",
)

Notebook helpers are available from saklas.notebook after installing saklas[notebook]: plot_alpha_sweep, plot_probe_correlation, plot_layer_norms, plot_trait_history, and to_dataframe.

Model support

Saklas has end-to-end tested paths for:

• Qwen 2, Qwen 3, and Qwen 3.5, including supported text and MoE variants;
• Gemma 2, Gemma 3, and Gemma 4, including text-only extraction from supported multimodal checkpoints;
• Mistral 3 and Ministral 3;
• Llama, GLM, gpt-oss, and Talkie.

Additional architectures are wired through the generic residual-layer interface, including Mixtral, Phi, Cohere, DeepSeek, OLMo, Granite, Nemotron, GPT-2-family, Falcon, MPT, DBRX, OPT, and others. Saklas emits a warning when an architecture is wired but has not been exercised end to end.

CUDA and Apple Silicon MPS both have real-model smoke coverage. Model-specific features still depend on the checkpoint: chat/role experiments require a compatible chat template, official SAE and J-LENS sources cover only some models, and FlashAttention depends on the model's Transformers attention implementation.

State and distribution

Saklas keeps local state under ~/.saklas/; set $SAKLAS_HOME to move it. The store contains authored manifolds, per-model fits and neutral statistics, local SAE/J-LENS artifacts, source bindings, and templates. Conversation saves are explicit browser-downloaded JSON files (or caller-selected LoomTree.save() paths); they are not autosaved under ~/.saklas/.

Manifold packs are folders with metadata, node corpora, integrity hashes, and optional fitted tensors. They can be installed from a local path or distributed as Hugging Face model repositories. A fitted two-node PCA manifold can also be exported as a llama.cpp control-vector GGUF with saklas pack export gguf.

Treat model repositories and downloaded artifacts as executable or otherwise untrusted input. Saklas verifies declared artifact hashes, but integrity is not authorship.

Development

pip install -e ".[dev]"

ruff check .
pyright
pytest -q -m "not gpu"
pytest -q tests/
python -m build

GPU smoke tests may download model weights:

pytest -q tests/test_smoke.py

The WebUI is a Svelte 5 + Vite application in webui/. Its compiled bundle under saklas/web/dist/ is committed package data and ships in the wheel.

cd webui
npm ci
npm run check
npm run build
git diff --exit-code ../saklas/web/dist

See CONTRIBUTING.md for development conventions and adding a model architecture.

Research lineage and credits

Saklas builds on Representation Engineering (Zou et al., 2023). repeng by Theia Vogel is the best-known compact implementation of that approach; Saklas takes the workbench route, adding live monitoring, manifold geometry, branching experiments, and server protocols.

Two-pole extraction uses difference-of-means following Im & Li, 2025. Manifold steering follows Goodfire's manifold work. The personas source is derived from the framing in Anthropic's Assistant Axis paper. J-LENS support implements the verbalizable-workspace method of Gurnee et al., 2026.

If you use Saklas in published research, please cite the relevant upstream methods alongside the Saklas version and exact model checkpoint you used.

Issues and security

Please update to the latest Saklas release before filing a bug. Include the model ID, device, dtype or quantization mode, Saklas version, and a minimal reproduction in GitHub Issues.

Report vulnerabilities privately according to SECURITY.md.

License

Saklas is licensed under AGPL-3.0-or-later. See LICENSE.