Fix README images
This commit is contained in:
parent
522a491fc1
commit
f7e8b1097b
1 changed files with 5 additions and 5 deletions
10
README.md
10
README.md
|
|
@ -10,7 +10,7 @@
|
|||
[](https://www.npmjs.com/package/@vectorize-io/hindsight-client)
|
||||
[](https://join.slack.com/t/hindsight-space/shared_invite/zt-3klo21kua-VUCC_zHP5rIcXFB1_5yw6A)
|
||||
|
||||
[Documentation](https://vectorize-io.github.io/hindsight) • [Paper](./Hindsight.pdf) • [Examples](./examples)
|
||||
[Documentation](https://vectorize-io.github.io/hindsight) • [Paper](#coming-soon) • [Examples](https://github.com/vectorize-io/hindsight-cookbook)
|
||||
|
||||
</div>
|
||||
|
||||
|
|
@ -28,7 +28,7 @@ Hindsight addresses common challenges that have frustrated AI engineers building
|
|||
|
||||
## How Hindsight Works
|
||||
|
||||

|
||||

|
||||
|
||||
Hindsight organizes memory into four networks to mimic the way human memory works:
|
||||
|
||||
|
|
@ -156,7 +156,7 @@ client.retain(
|
|||
|
||||
Behind the scenes, the retain operation uses an LLM to extract key facts, temporal data, entities, and relationships. It passes these through a normalization process to transform extracted data into canonical entities, time series, and search indexes along with metadata. These representations create the pathways for accurate memory retrieval in the recall and reflect operations.
|
||||
|
||||

|
||||

|
||||
|
||||
### Recall
|
||||
|
||||
|
|
@ -182,7 +182,7 @@ Recall performs 4 retrieval strategies in parallel:
|
|||
- Graph: Entity/temporal/causal links
|
||||
- Temporal: Time range filtering
|
||||
|
||||

|
||||

|
||||
|
||||
The individual results from the retrievals are merged, then ordered by relevance using reciprocal rank fusion and a cross-encoder reranking model.
|
||||
|
||||
|
|
@ -208,7 +208,7 @@ client = Hindsight(base_url="http://localhost:8888")
|
|||
client.reflect(bank_id="my-bank", query="What should I know about Alice?")
|
||||
```
|
||||
|
||||

|
||||

|
||||
|
||||
## Integrations
|
||||
|
||||
|
|
|
|||
Loading…
Reference in a new issue