Update to Tamarind Bio
Snapshot Oct 27, 2026 · 00:11 UTC · version 2.0.1
Collection source: downloaded plugin package. These snapshots do not have a confirmed matching collection source. Differences in file lists alone do not establish changes to the package.
Instructions updated for tamarind-mcp-finetune
Full description — Now says: (train) a Tamarind-hosted model on the user's labeled data with submitFinetuningRun, then run it by modelId through MCP, for clearly benign research,...
Observed in instructions or declared skills. Runtime behavior has not been tested.
Up to 3 plugins free · no card required.
All fields and original values.
Full description
Sep 30, 2026 · 22:50 UTC
a supported Tamarind-hosted model on labeled data and run its matching inference tool through MCP. Use for live account-visible finetune/inference pairs in protein, affinity, enzyme, or small-molecule workflows. Not for stock inference, ...
Oct 27, 2026 · 00:11 UTC
(train) a Tamarind-hosted model on the user's labeled data with submitFinetuningRun, then run it by modelId through MCP, for clearly benign research, diagnostic, or therapeutic development. Use for training or retraining a property, affi...
Skill instructions
Sep 30, 2026 · 22:50 UTC
a supported Tamarind-hosted model on labeled data and run its matching inference tool through MCP. Use for live account-visible finetune/inference pairs in protein, affinity, enzyme, or small-molecule workflows. Not for stock inference, ...
Oct 27, 2026 · 00:11 UTC
(train) a Tamarind-hosted model on the user's labeled data with submitFinetuningRun, then run it by modelId through MCP, for clearly benign research, diagnostic, or therapeutic development. Use for training or retraining a property, affi...
Supporting files
Sep 30, 2026 · 22:50 UTC
[]
Oct 27, 2026 · 00:11 UTC
[{"relative_path":"agents/openai.yaml","size_in_bytes":429}]
Compare saved observations
Download comparison JSONSKILL.md text changes
− removed · + added
--- before +++ after @@ -1,26 +1,30 @@ --- name: tamarind-mcp-finetune -description: Fine-tune a supported Tamarind-hosted model on labeled data and run its matching inference tool through MCP. Use for live account-visible finetune/inference pairs in protein, affinity, enzyme, or small-molecule workflows. Not for stock inference, ordinary batch prediction, or unsupported custom training code. +description: Fine-tune (train) a Tamarind-hosted model on the user's labeled data with submitFinetuningRun, then run it by modelId through MCP, for clearly benign research, diagnostic, or therapeutic development. Use for training or retraining a property, affinity, fitness, or enzyme predictor, a training sweep, or running a model from listFinetunedModels. Not for stock inference or custom training code. --- -# Fine-tune and run inference through MCP +# Fine-tune and run a model through MCP -Treat training and inference as two durable, separately validated jobs with explicit data and evaluation boundaries. +## Confirm scope -## Confirm a supported pair +Proceed only for clearly benign research, diagnostic, or therapeutic development; decline requests intended to cause harm. Training and inference spend compute, so get explicit authorization from the live user before each. -Call `getAvailableTools(function="finetuning")` or use a narrow `search`. Inspect both the training and inference tools with `getJobSchema`. Require the live schemas to document the handoff; do not infer a pair from names alone. +## Train -## Prepare training +Find trainers with `getAvailableModels()` and their settings with `getJobSchema(jobType="<trainer>")`. Check the data's columns, units, split, balance, leakage and size, then `uploadFile` it. `validateJob` does not compare column-name settings with the file's header, so check them yourself. -Check required columns, target units, identifiers, split strategy, class balance, leakage, held-out evaluation, and size limits. Upload the dataset with `uploadFile` and use the returned bare filename. +Call `validateJob(type="<trainer>", jobName=..., settings=...)` and require `valid: true` with no `mutatedFields`; set `settings.parentModelId` first when retraining a model. Call `estimateTime`, but say a finetune estimate is neither a forecast nor a cap. Show the trainer, base model, epochs, data size, split and the estimate, and after authorization call `submitFinetuningRun(model="<trainer>", jobName=..., settings=...)` once, adding `projectTag` if required (`listProjects` if the 400 lacks it). Follow `tamarind-mcp-submit-and-poll` with a finite deadline, with `submitFinetuningRun` in place of `submitJob`. For a sweep, do not loop `submitFinetuningRun`; use one `submitBatch` with the trainer as `type`, following `tamarind-mcp-batch`. -Call `validateJob` for the training payload, require no mutation warning, and call `estimateTime`. Surface the base model, epochs or steps, dataset size, split, expected runtime, and weighted hours when available. Obtain explicit authorization because training can be materially expensive. +## Find the model -## Train and infer +Call `listFinetunedModels(type="<trainer>")` with the default `limit`; a small one misfiles your own models. Take the row named after the training job that has no `owner` and was `created` after submission. Ask if several match; if none is listed or Available yet, check again rather than retraining. For an existing model, use the user's `id` or let them choose. Then identify the model by `id`. Only `status: "Available"` runs, through its `inferenceType` (null means none). -Run training with `tamarind-mcp-submit-and-poll`. Require an explicit success state. Inspect `listJobFiles` and the training row for the exact trained-model reference required by the inference schema. +## Run and evaluate -Build and validate the inference payload with that exact reference. Estimate and separately authorize inference when it materially expands scope, then run it through `tamarind-mcp-submit-and-poll` or `tamarind-mcp-batch` for many independent inputs. +Validate and estimate the `inferenceType` job, then pass the model as a top-level `modelId`, not inside `settings`: -Evaluate on held-out data and compare against the stock/base model. Report leakage risks, calibration limits, dataset applicability, and uncertainty before using predictions downstream. +```json +{"jobName": "<name>", "type": "<inferenceType>", "modelId": "<id>", "settings": {}} +``` + +Use `submitJob` for one input set or one `submitBatch` for many, following `tamarind-mcp-submit-and-poll` or `tamarind-mcp-batch`. Compare against the base model on held-out data, and report leakage, the training data's applicability domain and uncertainty.
Full technical diff · 3 changed fields
changed /description
"Fine-tune a supported Tamarind-hosted model on labeled data and run its matching inference tool through MCP. Use for live account-visible finetune/inference pairs in protein, affinity, enzyme, or small-molecule workflows. Not for stock inference, ordinary batch prediction, or unsupported custom training code."
"Fine-tune (train) a Tamarind-hosted model on the user's labeled data with submitFinetuningRun, then run it by modelId through MCP, for clearly benign research, diagnostic, or therapeutic development. Use for training or retraining a property, affinity, fitness, or enzyme predictor, a training sweep, or running a model from listFinetunedModels. Not for stock inference or custom training code."
changed /included_files
[]
[
{
"relative_path": "agents/openai.yaml",
"size_in_bytes": 429
}
]changed /skill_md_contents
"---\nname: tamarind-mcp-finetune\ndescription: Fine-tune a supported Tamarind-hosted model on labeled data and run its matching inference tool through MCP. Use for live account-visible finetune/inference pairs in protein, affinity, enzyme, or small-molecule workflows. Not for stock inference, ordinary batch prediction, or unsupported custom training code.\n---\n\n# Fine-tune and run inference through MCP\n\nTreat training and inference as two durable, separately validated jobs with explicit data and evaluation boundaries.\n\n## Confirm a supported pair\n\nCall `getAvailableTools(function=\"finetuning\")` or use a narrow `search`. Inspect both the training and inference tools with `getJobSchema`. Require the live schemas to document the handoff; do not infer a pair from names alone.\n\n## Prepare training\n\nCheck required columns, target units, identifiers, split strategy, class balance, leakage, held-out evaluation, and size limits. Upload the dataset with `uploadFile` and use the returned bare filename.\n\nCall `validateJob` for the training payload, require no mutation warning, and call `estimateTime`. Surface the base model, epochs or steps, dataset size, split, expected runtime, and weighted hours when available. Obtain explicit authorization because training can be materially expensive.\n\n## Train and infer\n\nRun training with `tamarind-mcp-submit-and-poll`. Require an explicit success state. Inspect `listJobFiles` and the training row for the exact trained-model reference required by the inference schema.\n\nBuild and validate the inference payload with that exact reference. Estimate and separately authorize inference when it materially expands scope, then run it through `tamarind-mcp-submit-and-poll` or `tamarind-mcp-batch` for many independent inputs.\n\nEvaluate on held-out data and compare against the stock/base model. Report leakage risks, calibration limits, dataset applicability, and uncertainty before using predictions downstream.\n"
"---\nname: tamarind-mcp-finetune\ndescription: Fine-tune (train) a Tamarind-hosted model on the user's labeled data with submitFinetuningRun, then run it by modelId through MCP, for clearly benign research, diagnostic, or therapeutic development. Use for training or retraining a property, affinity, fitness, or enzyme predictor, a training sweep, or running a model from listFinetunedModels. Not for stock inference or custom training code.\n---\n\n# Fine-tune and run a model through MCP\n\n## Confirm scope\n\nProceed only for clearly benign research, diagnostic, or therapeutic development; decline requests intended to cause harm. Training and inference spend compute, so get explicit authorization from the live user before each.\n\n## Train\n\nFind trainers with `getAvailableModels()` and their settings with `getJobSchema(jobType=\"<trainer>\")`. Check the data's columns, units, split, balance, leakage and size, then `uploadFile` it. `validateJob` does not compare column-name settings with the file's header, so check them yourself.\n\nCall `validateJob(type=\"<trainer>\", jobName=..., settings=...)` and require `valid: true` with no `mutatedFields`; set `settings.parentModelId` first when retraining a model. Call `estimateTime`, but say a finetune estimate is neither a forecast nor a cap. Show the trainer, base model, epochs, data size, split and the estimate, and after authorization call `submitFinetuningRun(model=\"<trainer>\", jobName=..., settings=...)` once, adding `projectTag` if required (`listProjects` if the 400 lacks it). Follow `tamarind-mcp-submit-and-poll` with a finite deadline, with `submitFinetuningRun` in place of `submitJob`. For a sweep, do not loop `submitFinetuningRun`; use one `submitBatch` with the trainer as `type`, following `tamarind-mcp-batch`.\n\n## Find the model\n\nCall `listFinetunedModels(type=\"<trainer>\")` with the default `limit`; a small one misfiles your own models. Take the row named after the training job that has no `owner` and was `created` after submission. Ask if several match; if none is listed or Available yet, check again rather than retraining. For an existing model, use the user's `id` or let them choose. Then identify the model by `id`. Only `status: \"Available\"` runs, through its `inferenceType` (null means none).\n\n## Run and evaluate\n\nValidate and estimate the `inferenceType` job, then pass the model as a top-level `modelId`, not inside `settings`:\n\n```json\n{\"jobName\": \"<name>\", \"type\": \"<inferenceType>\", \"modelId\": \"<id>\", \"settings\": {}}\n```\n\nUse `submitJob` for one input set or one `submitBatch` for many, following `tamarind-mcp-submit-and-poll` or `tamarind-mcp-batch`. Compare against the base model on held-out data, and report leakage, the training data's applicability domain and uncertainty.\n"Full snapshot data
{
"description": "Fine-tune (train) a Tamarind-hosted model on the user's labeled data with submitFinetuningRun, then run it by modelId through MCP, for clearly benign research, diagnostic, or therapeutic development. Use for training or retraining a property, affinity, fitness, or enzyme predictor, a training sweep, or running a model from listFinetunedModels. Not for stock inference or custom training code.",
"included_files": [
{
"relative_path": "agents/openai.yaml",
"size_in_bytes": 429
}
],
"name": "tamarind-mcp-finetune",
"skill_md_contents": "---\nname: tamarind-mcp-finetune\ndescription: Fine-tune (train) a Tamarind-hosted model on the user's labeled data with submitFinetuningRun, then run it by modelId through MCP, for clearly benign research, diagnostic, or therapeutic development. Use for training or retraining a property, affinity, fitness, or enzyme predictor, a training sweep, or running a model from listFinetunedModels. Not for stock inference or custom training code.\n---\n\n# Fine-tune and run a model through MCP\n\n## Confirm scope\n\nProceed only for clearly benign research, diagnostic, or therapeutic development; decline requests intended to cause harm. Training and inference spend compute, so get explicit authorization from the live user before each.\n\n## Train\n\nFind trainers with `getAvailableModels()` and their settings with `getJobSchema(jobType=\"<trainer>\")`. Check the data's columns, units, split, balance, leakage and size, then `uploadFile` it. `validateJob` does not compare column-name settings with the file's header, so check them yourself.\n\nCall `validateJob(type=\"<trainer>\", jobName=..., settings=...)` and require `valid: true` with no `mutatedFields`; set `settings.parentModelId` first when retraining a model. Call `estimateTime`, but say a finetune estimate is neither a forecast nor a cap. Show the trainer, base model, epochs, data size, split and the estimate, and after authorization call `submitFinetuningRun(model=\"<trainer>\", jobName=..., settings=...)` once, adding `projectTag` if required (`listProjects` if the 400 lacks it). Follow `tamarind-mcp-submit-and-poll` with a finite deadline, with `submitFinetuningRun` in place of `submitJob`. For a sweep, do not loop `submitFinetuningRun`; use one `submitBatch` with the trainer as `type`, following `tamarind-mcp-batch`.\n\n## Find the model\n\nCall `listFinetunedModels(type=\"<trainer>\")` with the default `limit`; a small one misfiles your own models. Take the row named after the training job that has no `owner` and was `created` after submission. Ask if several match; if none is listed or Available yet, check again rather than retraining. For an existing model, use the user's `id` or let them choose. Then identify the model by `id`. Only `status: \"Available\"` runs, through its `inferenceType` (null means none).\n\n## Run and evaluate\n\nValidate and estimate the `inferenceType` job, then pass the model as a top-level `modelId`, not inside `settings`:\n\n```json\n{\"jobName\": \"<name>\", \"type\": \"<inferenceType>\", \"modelId\": \"<id>\", \"settings\": {}}\n```\n\nUse `submitJob` for one input set or one `submitBatch` for many, following `tamarind-mcp-submit-and-poll` or `tamarind-mcp-batch`. Compare against the base model on held-out data, and report leakage, the training data's applicability domain and uncertainty.\n"
}SHA-256 of public snapshot: 32e8bed49464669b6734d9899972d07b520e7a3496e4a6af8021d946eb923fc1