SKILLEMALL.ai
GitHub ↗ RU Sign in

CD train-sentence-transformers

Train or fine-tune SentenceTransformer, CrossEncoder, and SparseEncoder models for retrieval, similarity, clustering, classification, reranking, and related embedding tasks.

sickn33/agentic-awesome-skills Agent Skills author: sickn33 MIT 32 files · 15 scripts body ≈ 2 507 tokens Open the sourcegithub.com↗ analyzed 5 d ago

Train or fine-tune SentenceTransformer, CrossEncoder, and SparseEncoder models for retrieval, similarity, clustering, classification, reranking, and related…

As a process D 44/100 · Unfinished process — References files that are not bundled: scripts/train_<type>_example.py, scripts/train_sentence_transformer_<matryoshka|multi_dataset|with_lora|distillation|make_multilingual|static_embedding>_example.py, scripts/train_cross_encoder_<distillation|listwise>_example.py

GeneratorAI and agentstype and topics are labelled automatically from the skill text
Runs in: Claude / Claude Code ChatGPT / Codex Gemini CLI Cursor GitHub Copilot / VS Code Qwen Code Kimi Code CLI Z.ai (GLM) DeepSeek (Deep Code) Mistral Vibe LangChain (deepagents) OpenClaw Goose OpenCode
JSON
Technical rating
C
84/100
safety, quality, tests
Safety 60%
93
Quality 40%
71
Run on models
none yet
Process rating
D
44/100
Unfinished process
References files that are not bundled: scripts/train_<type>_example.py, scripts/train_sentence_transformer_<matryoshka|multi_dataset|with_lora|distillation|make_multilingual|static_embedding>_example.py, scripts/train_cross_encoder_<distillation|listwise>_example.py
Tools and files w 18
0
Failures and branches w 10
0
Progress reporting w 2
0
the three weakest of ten parameters · all ten

The same skill appears in 2 more places: agentic-awesome-skills, agentic-awesome-skills

How to improve

  1. The text references files that are not there: add them or drop the references.
For the model run — optional
  • Your own cases (evals/evals.json, 4–6 real requests with expected answers): the full check would then run those instead of a model-drafted suite.
  • A spec.yaml with trigger phrases and assertions — a behaviour contract for CI; `skilltest init` writes a template.

Guard findings · 7

✓ No critical or high findings

Medium and low: 7
  • low Secrets in code secret-high-entropy-token references/base_model_selection.md:44
    High-entropy token-like string (may be an id, hash or a credential) (quoted — discussed, not commanded)
    - **CPU / small footprint** (`StaticEmbedding`): `StaticEmbedding(tokenizer, embedding_dim=...)`. **Model size = `vocab_size × dim × 4 bytes`** — pick a small-vocab tokenizer or you get a giant model:
    quoted
  • low Secrets in code secret-high-entropy-token references/evaluators_cross_encoder.md:36
    High-entropy token-like string (may be an id, hash or a credential) (quoted — discussed, not commanded)
    Output key for `metric_for_best_model`: **`eval…dcg@10`**. The `R100` signals "rerank top-100"; if you change `rerank_k`, the prefix changes (e.g. `R50`).
    quoted
  • low Secrets in code secret-high-entropy-token references/evaluators_cross_encoder.md:106
    High-entropy token-like string (may be an id, hash or a credential) (quoted — discussed, not commanded)
    - `eval…dcg@10` — `CrossEncoderNanoBEIREvaluator` default
    quoted
  • low Secrets in code secret-high-entropy-token references/losses_cross_encoder.md:101
    High-entropy token-like string (may be an id, hash or a credential) (placeholder value)
    - **Tiny loss at large K is expected.** With `NDCG…eme`, the loss normalizes by the discount-weighted pair count — at K=128 the loss can scale to ~`1e-4` numerically. That's not "training bro
    placeholder
  • low Secrets in code secret-high-entropy-token references/training_args.md:148
    High-entropy token-like string (may be an id, hash or a credential) (quoted — discussed, not commanded)
    - `CrossEncoderNanoBEIREvaluator` (rerank from BM25 top-100): `eval…dcg@10`
    quoted
  • low Secrets in code secret-high-entropy-token references/troubleshooting.md:148
    High-entropy token-like string (may be an id, hash or a credential) (placeholder value)
    **Fix:** ignore training loss for LambdaLoss; watch the eval metric (`eval…dcg@10` or your `metric_for_best_model`) instead. If eval is moving in the right direction and loss is "t
    placeholder
  • low Secrets in code secret-high-entropy-token scripts/train_sentence_transformer_make_multilingual_example.py:32
    High-entropy token-like string (may be an id, hash or a credential) (test fixture / example file; quoted — discussed, not commanded)
    - Student: must be multilingual (xlm-roberta-base, para…-v2,
    fixturequoted

Files scanned: 31. Evidence is masked. Grey chips explain why severity was lowered.

Against the Agent Skills spec

  • warning missing-ref reference to a missing file: scripts/train_<type>_example.py
  • warning missing-ref reference to a missing file: scripts/train_sentence_transformer_<matryoshka|multi_dataset|with_lora|distillation|make_multilingual|static_embedding>_example.py
  • warning missing-ref reference to a missing file: scripts/train_cross_encoder_<distillation|listwise>_example.py
  • warning missing-ref reference to a missing file: references/losses_<type>.md
  • warning missing-ref reference to a missing file: references/evaluators_<type>.md
  • note frontmatter-key unknown frontmatter key "risk"
  • note frontmatter-key unknown frontmatter key "source"
  • note frontmatter-key unknown frontmatter key "source_repo"
  • note frontmatter-key unknown frontmatter key "source_type"
  • note frontmatter-key unknown frontmatter key "date_added"
  • note frontmatter-key unknown frontmatter key "license_source"
  • note edit-residue the text marks something as outdated (lines 46): check that old rules are not kept next to new ones — the full check reads the text for contradictions

Process rating: all ten parameters 44/100

Will not run. References files that are not bundled: scripts/train_<type>_example.py, scripts/train_sentence_transformer_<matryoshka|multi_dataset|with_lora|distillation|make_multilingual|static_embedding>_example.py, scripts/train_cross_encoder_<distillation|listwise>_example.py
  • 0Tools and files. 5 referenced file(s) missing: scripts/train_<type>_example.py, scripts/train_sentence_transformer_<matryoshka|multi_dataset|with_lora|distillation|make_multilingual|static_embedding>_example.py, scripts/train_cross_encoder_<distillation|listwise>_example.py
  • 0Failures and branches. Linear process with no failure handling
  • 0Progress reporting. Says nothing while it works
  • 20When it triggers. No condition that starts the skill
  • 30Running it twice. 7 mutating operations with no state check
  • 40Result and completion. Does not say what the result is
  • 70Inputs and preconditions. Inputs and preconditions are listed
  • 85Steps. 31 steps, 1 vague phrases
  • 100Consistency. Name and required fields are in place
  • 100Execution cost. Instruction body is 2507 tokens
  • low The skill ranks results itself: that belongs to the system behind the tool, not the model
  • low The response is described with custom markup (4 tags): a typed call is more reliable

Everything here is measured from the skill text rather than judged by a model, so the numbers are checkable. A parameter weighs more when it is a more common reason for the process to stall.

Quality signals

  • +5Description has no quoted example phrases that should trigger the skill
  • +4Description does not say when NOT to use the skill (false activations)
  • +3Output format is not stated: the model decides each time
  • -38 of 14 scripts are never mentioned in SKILL.md
  • +2Single-language instructions
  • +3Description length 173: enough signal without eating the budget
  • +4Structure: 13 headings
  • +3Step-by-step instructions: 31 items
  • +4Has examples (1 code blocks)
  • +4Reference files are cited in the instructions (14 of 14)
  • +1License stated

Quality base 70; lint remarks subtract, signals add up to 100. Result: 71.