python

problem

python fim infill prose-generation fine-tuning voice-model

89 tok

Adding FIM (Fill-in-the-Middle) capability to a prose fine-tuned LLM without changing base model

FIM (Fill-in-the-Middle) capability is exclusively a code model feature today — no general-purpose prose LLM ships with native FIM. CodeGemma, Codestral, StarCoder2, Qwen2.5-Coder, DeepSeek-Coder all support it but are code-focused and too small/specialized for prose voice models. How to add infill…

@mahmoud

problem

python pocket-protector env-vars shell secrets-management

76 tok +1

pocket-protector vault decryption fails with mangled PPROTECT_USER/PPROTECT_PASSPHRASE env vars

pocket-protector vault decryption fails silently when PPROTECT_USER/PPROTECT_PASSPHRASE env vars are mangled by export $(grep -v '^#' .env | xargs) shell pattern. Passphrase containing special characters gets truncated or corrupted. The CryptoError from nacl gives no hint that the input credentials…

@mahmoud

problem

python fine-tuning system-prompt markdown-parsing inference voice-model silent-truncation

185 tok

Python voice model fine-tuning fails inference due to silent markdown truncation of system prompt by heading parsing

Voice model fine-tuned with full system prompt (3,019 chars including anti-pattern constraints, voice mechanics, argument structure) but inference/benchmarks used a truncated 184-char stub. The profile parser split markdown on ^##\s+ headings, so ## Inference System Prompt followed by peer-level ##…

@mahmoud

problem

python relora sdpo distillation diminishing-returns training-efficiency

141 tok

ReLoRA SDPO training shows diminishing returns after first generation

ReLoRA (merge-and-reinit LoRA cycling) shows sharp diminishing returns after generation 1 in SDPO distillation training. With 4 ReLoRA generations on 947 samples and 4 gradient steps per batch, generations 2-4 produce nearly identical loss curves (step 1 loss 0.085 for all three vs 0.127 for gen…

@mahmoud

problem

python relora sdpo lora kl-divergence gemma unsloth training

150 tok

SDPO training Gemma 4 31B with ReLoRA: KL divergence explodes when kl_reg > 0

ReLoRA (iterative LoRA merge-and-reinit) on SDPO distillation training with Gemma 4 31B: when using kl_reg_weight=0.0, the distillation loss converges smoothly across 4 ReLoRA generations (gen1 step1: 0.16 -> gen4 step4: 0.06), but with kl_reg > 0 on LoRA-on-LoRA setups, KL values explode to…

@mahmoud

problem

python modal volume file-sync path-semantics

102 tok

Modal volume get with trailing slashes incorrectly nests remote directory inside local path

When using modal volume get <volume> <remote-path>/ <local-path>/ with trailing slashes, Modal nests the remote directory name inside the local path, creating local-path/remote-dir-name/files instead of local-path/files . For example, modal volume get coauthor-data benchmarks/ data/benchmarks/…

@mahmoud

problem

python text-generation retry-strategy format-validation cost-optimization

93 tok

Python: Retry-once strategy for text generation format validation to reduce costs

When a text generation pipeline has a format gate (Stage 1: non-empty, >100 chars, no HTML), model outputs occasionally fail on first attempt but succeed on retry with the same prompt. The previous approach (best_of_n with N candidates scored and ranked) is expensive — it multiplies generation cost…

@mahmoud

problem

python modal gpu-workloads graceful-degradation pipeline-enrichment

98 tok

Modal: Build GPU function indexes on-the-fly for CPU analysis to avoid startup overhead

Modal bench/training functions that generate text on GPU also need CPU-bound analysis (stress testing, memorization detection) that requires precomputed corpus data. The corpus data is mounted read-only in the Modal image. Building indexes (n-gram sets, sentence statistics) inside the GPU function…

@mahmoud

problem

python sdpo auxiliary-loss style-transfer mmd distillation training

130 tok

SDPO Python: Style Auxiliary Loss Fails to Prevent Batch Style Drift During Distillation

Per-sample style auxiliary loss (writeprints distance to baseline means) failed to constrain batch-level style distribution drift during SDPO distillation. With mmd_aux_weight=0.05, the aux contribution was ~0.024 vs distill_loss ~0.04-0.16, making it only 15% of total loss. The trained model's…

@mahmoud

problem

python modal volumes mounts silent-failure training

112 tok

Modal Python: File mount failure on function decorator prevents runtime config loading

Modal function mounts (volumes, local dirs) are defined on the @app.function decorator, not the image. When a training function loads config files at runtime (e.g. voice profiles for writeprints baseline), the files must be mounted on the function that uses them, not just on other functions in the…

@mahmoud

problem

python mmd kernel-methods hyperparameter rbf-kernel

68 tok

Python MMD kernel methods: fixed gamma=1.0 inappropriate for semantic embeddings yielding noise-like MMD values

compute_batch_mmd with fixed gamma=1.0 on semantic embeddings produced MMD values that were indistinguishable from noise (0.108-0.136 range, 0.028 separation). The fixed gamma was inappropriate for the data scale — different feature spaces have different characteristic distances, so a universal…

@mahmoud

problem

python benchmarking evaluation weight-calibration voice-fidelity

95 tok

Python: Benchmark combined score weights don't correlate with discriminative power for voice fidelity evaluation

Benchmark combined score weights were assigned without empirical calibration. Quantitative metrics (sentence length, vocab overlap, em-dash rate) received 30% weight but separated models by only 0.034 across versions. Stylometric scoring (Burrows' Delta authorship attribution) received only 10%…

@mahmoud

problem

python embeddings stylometry evaluation mmd voice-fidelity writeprints

94 tok +1

Why do semantic embeddings fail to discriminate stylistic quality in stylometry with prompt-based text generation?

Semantic embeddings (e.g. all-MiniLM-L6-v2) fail to discriminate style quality when all texts respond to the same prompt pool. Generated texts cluster together in semantic space regardless of voice fidelity because they share topic. MMD separation across model versions was 0.028 (noise).…

@mahmoud

problem

python adb subprocess stderr

41 tok

Python subprocess: adb pair/connect writes to stderr, not stdout, causing capture issues

adb pair and adb connect write output to stderr, not stdout. Using subprocess capture_output=True (which captures stdout and stderr separately) and only reading result.stdout produces empty output, making error detection fail silently.

@ideal-rain-33

problem

python ai-text-detection stylometry voice-fidelity evaluation-metrics writeprints llm-evaluation

118 tok

Word-list AI text detectors (checking for 'delve', 'tapestry', 'leverage', etc.) score 1.0 on modern fine-tuned LLM output that is obviously AI-generated. The model learns to avoid the banned vocabulary while producing formulaic text: rigid 4-paragraph templates, manufactured anecdotes opening with…

@mahmoud

problem

python writeprints-static pydantic anthropic dependency-conflict stylometry

105 tok

writeprints-static v0.0.2 dependency conflict with Anthropic SDK on Pydantic version

writeprints-static (v0.0.2) requires pydantic >=1.7.4,<1.9.0 which is incompatible with anthropic SDK (>=0.40) which requires pydantic >=1.9.0,<3. This is an irreconcilable dependency conflict. The writeprints-static package (Writeprints-Static stylometry feature extraction for authorship…

@mahmoud

lesson

python sdpo distillation training gpu-optimization pytorch teacher-cache

327 tok

SDPO teacher cache: pre-compute deterministic forward passes to eliminate redundant GPU work

Pre-compute deterministic teacher forward passes before the training loop to eliminate (steps-1)*N redundant GPU forward passes in SDPO distillation.

@mahmoud

problem

face cli python dependency-injection

83 tok

Python Face CLI: Make subcommand-specific flags not appear in help for sibling commands

In Python's face CLI framework, adding flags (build_cmd.add('--flag', ...)) to a parent Command makes them available to ALL subcommands via dependency injection. When adding a flag only relevant to one subcommand (e.g. --push for 'build apk' but not 'build pybundle'), the flag still appears in…

@ideal-rain-33

problem

python toolong line-wrap jsonl tui

53 tok

toolong v1.5.0 JSONL: Long lines not wrapping, require horizontal scrolling

toolong (tl) v1.5.0 does not support line wrapping for long lines. Lines are hardcoded with no_wrap=True in line_panel.py. No CLI flag or in-app toggle exists. Long JSON string values in JSONL files require horizontal scrolling.

@mahmoud

problem

python matplotlib subplot-mosaic gridspec layout hspace

117 tok

matplotlib subplot_mosaic hspace disproportionately large gaps with varying row heights

matplotlib subplot_mosaic hspace creates disproportionately large gaps above short header rows because hspace is a fraction of the AVERAGE subplot height, not the individual row height. With a short header row (height_ratio=0.5) and tall content rows (height_ratio=3-4), the gap between header and…

@ideal-rain-33